diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,147360 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.999872737916466,
+  "eval_steps": 500,
+  "global_step": 24555,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219512e-06,
+      "loss": 0.8901,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4390243902439023e-06,
+      "loss": 0.8157,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6585365853658533e-06,
+      "loss": 0.8273,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-06,
+      "loss": 0.8642,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.097560975609756e-06,
+      "loss": 0.9531,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3170731707317065e-06,
+      "loss": 0.7688,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.536585365853657e-06,
+      "loss": 0.9389,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75609756097561e-06,
+      "loss": 0.8463,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0975609756097559e-05,
+      "loss": 0.8358,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219511e-05,
+      "loss": 0.7671,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3414634146341462e-05,
+      "loss": 0.7806,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4634146341463413e-05,
+      "loss": 0.8797,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5853658536585366e-05,
+      "loss": 0.6765,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7073170731707313e-05,
+      "loss": 0.8248,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8292682926829268e-05,
+      "loss": 0.7479,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.951219512195122e-05,
+      "loss": 0.9234,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.073170731707317e-05,
+      "loss": 0.743,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1951219512195117e-05,
+      "loss": 0.7324,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.317073170731707e-05,
+      "loss": 0.6685,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4390243902439022e-05,
+      "loss": 0.6609,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5609756097560977e-05,
+      "loss": 0.5058,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6829268292682924e-05,
+      "loss": 0.5153,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8048780487804875e-05,
+      "loss": 0.511,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9268292682926826e-05,
+      "loss": 0.4987,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0487804878048777e-05,
+      "loss": 0.4468,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.4207,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.292682926829268e-05,
+      "loss": 0.3482,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4146341463414627e-05,
+      "loss": 0.2525,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5365853658536584e-05,
+      "loss": 0.2636,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 0.2023,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.780487804878048e-05,
+      "loss": 0.2962,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.902439024390244e-05,
+      "loss": 0.2923,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.024390243902439e-05,
+      "loss": 0.2076,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.146341463414634e-05,
+      "loss": 0.2102,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.268292682926829e-05,
+      "loss": 0.1848,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3902439024390234e-05,
+      "loss": 0.2199,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.512195121951219e-05,
+      "loss": 0.1717,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.634146341463414e-05,
+      "loss": 0.2331,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7560975609756094e-05,
+      "loss": 0.2204,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8780487804878045e-05,
+      "loss": 0.137,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 0.1524,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.121951219512195e-05,
+      "loss": 0.1477,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.24390243902439e-05,
+      "loss": 0.2392,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.365853658536585e-05,
+      "loss": 0.1178,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.48780487804878e-05,
+      "loss": 0.172,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.609756097560975e-05,
+      "loss": 0.1596,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.731707317073171e-05,
+      "loss": 0.1053,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.853658536585365e-05,
+      "loss": 0.181,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.97560975609756e-05,
+      "loss": 0.1285,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0975609756097554e-05,
+      "loss": 0.2609,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.21951219512195e-05,
+      "loss": 0.0859,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.341463414634146e-05,
+      "loss": 0.1277,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.463414634146341e-05,
+      "loss": 0.0935,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.585365853658536e-05,
+      "loss": 0.148,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.707317073170731e-05,
+      "loss": 0.0975,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.829268292682925e-05,
+      "loss": 0.1179,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.951219512195121e-05,
+      "loss": 0.0992,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.073170731707317e-05,
+      "loss": 0.1607,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.195121951219511e-05,
+      "loss": 0.1898,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.317073170731707e-05,
+      "loss": 0.0832,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.439024390243901e-05,
+      "loss": 0.1476,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.560975609756096e-05,
+      "loss": 0.1459,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.682926829268292e-05,
+      "loss": 0.1631,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.804878048780487e-05,
+      "loss": 0.1454,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.926829268292682e-05,
+      "loss": 0.1268,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.048780487804878e-05,
+      "loss": 0.1435,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.170731707317072e-05,
+      "loss": 0.1703,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.292682926829268e-05,
+      "loss": 0.1395,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.414634146341464e-05,
+      "loss": 0.1493,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.536585365853658e-05,
+      "loss": 0.0931,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.658536585365854e-05,
+      "loss": 0.1304,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.780487804878047e-05,
+      "loss": 0.0573,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.902439024390243e-05,
+      "loss": 0.1005,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.024390243902438e-05,
+      "loss": 0.1031,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.146341463414633e-05,
+      "loss": 0.1224,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.268292682926829e-05,
+      "loss": 0.1241,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.390243902439023e-05,
+      "loss": 0.1499,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.512195121951219e-05,
+      "loss": 0.065,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.634146341463415e-05,
+      "loss": 0.0461,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.756097560975609e-05,
+      "loss": 0.1377,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.878048780487805e-05,
+      "loss": 0.0652,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.0434,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010121951219512195,
+      "loss": 0.0873,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001024390243902439,
+      "loss": 0.057,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010365853658536584,
+      "loss": 0.1902,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001048780487804878,
+      "loss": 0.1914,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010609756097560974,
+      "loss": 0.1099,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001073170731707317,
+      "loss": 0.1579,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010853658536585365,
+      "loss": 0.1448,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001097560975609756,
+      "loss": 0.1048,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011097560975609756,
+      "loss": 0.0703,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001121951219512195,
+      "loss": 0.0778,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011341463414634146,
+      "loss": 0.2221,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011463414634146342,
+      "loss": 0.1015,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011585365853658535,
+      "loss": 0.1234,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001170731707317073,
+      "loss": 0.0951,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011829268292682925,
+      "loss": 0.0773,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001195121951219512,
+      "loss": 0.1571,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012073170731707316,
+      "loss": 0.0873,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012195121951219511,
+      "loss": 0.0332,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012317073170731708,
+      "loss": 0.0526,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000124390243902439,
+      "loss": 0.0885,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012560975609756097,
+      "loss": 0.1175,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012682926829268293,
+      "loss": 0.1182,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012804878048780486,
+      "loss": 0.0798,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012926829268292681,
+      "loss": 0.0659,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013048780487804877,
+      "loss": 0.0878,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013170731707317073,
+      "loss": 0.1082,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001329268292682927,
+      "loss": 0.1111,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013414634146341462,
+      "loss": 0.0777,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013536585365853658,
+      "loss": 0.0633,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001365853658536585,
+      "loss": 0.07,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013780487804878046,
+      "loss": 0.1092,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013902439024390242,
+      "loss": 0.0514,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014024390243902438,
+      "loss": 0.095,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014146341463414634,
+      "loss": 0.0424,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014268292682926827,
+      "loss": 0.037,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014390243902439023,
+      "loss": 0.0788,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014512195121951218,
+      "loss": 0.0757,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014634146341463414,
+      "loss": 0.0382,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001475609756097561,
+      "loss": 0.0711,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014878048780487803,
+      "loss": 0.038,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015,
+      "loss": 0.0515,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015121951219512192,
+      "loss": 0.049,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001524390243902439,
+      "loss": 0.0646,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015365853658536583,
+      "loss": 0.1073,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001548780487804878,
+      "loss": 0.0996,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015609756097560975,
+      "loss": 0.1134,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001573170731707317,
+      "loss": 0.1024,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015853658536585364,
+      "loss": 0.0958,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001597560975609756,
+      "loss": 0.0443,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016097560975609755,
+      "loss": 0.1178,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001621951219512195,
+      "loss": 0.082,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016341463414634144,
+      "loss": 0.0791,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016463414634146343,
+      "loss": 0.079,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016585365853658536,
+      "loss": 0.0666,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001670731707317073,
+      "loss": 0.0789,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016829268292682927,
+      "loss": 0.065,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001695121951219512,
+      "loss": 0.0548,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017073170731707316,
+      "loss": 0.0742,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001719512195121951,
+      "loss": 0.0796,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017317073170731708,
+      "loss": 0.1732,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000174390243902439,
+      "loss": 0.0393,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017560975609756094,
+      "loss": 0.0439,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017682926829268292,
+      "loss": 0.1158,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017804878048780485,
+      "loss": 0.0425,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001792682926829268,
+      "loss": 0.061,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018048780487804877,
+      "loss": 0.0722,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018170731707317073,
+      "loss": 0.0714,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018292682926829266,
+      "loss": 0.0603,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001841463414634146,
+      "loss": 0.0475,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018536585365853657,
+      "loss": 0.0675,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018658536585365853,
+      "loss": 0.0315,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018780487804878046,
+      "loss": 0.0354,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018902439024390244,
+      "loss": 0.0566,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019024390243902437,
+      "loss": 0.0679,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001914634146341463,
+      "loss": 0.1309,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001926829268292683,
+      "loss": 0.1161,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019390243902439022,
+      "loss": 0.0409,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019512195121951218,
+      "loss": 0.0717,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001963414634146341,
+      "loss": 0.0329,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001975609756097561,
+      "loss": 0.0549,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019878048780487802,
+      "loss": 0.0514,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 0.0611,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020121951219512194,
+      "loss": 0.057,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002024390243902439,
+      "loss": 0.0812,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020365853658536583,
+      "loss": 0.0629,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002048780487804878,
+      "loss": 0.0566,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020609756097560974,
+      "loss": 0.0521,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020731707317073167,
+      "loss": 0.0665,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020853658536585363,
+      "loss": 0.033,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002097560975609756,
+      "loss": 0.0132,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021097560975609755,
+      "loss": 0.0796,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021219512195121948,
+      "loss": 0.0408,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021341463414634146,
+      "loss": 0.0795,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002146341463414634,
+      "loss": 0.1066,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021585365853658535,
+      "loss": 0.0334,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002170731707317073,
+      "loss": 0.0482,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021829268292682927,
+      "loss": 0.0386,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002195121951219512,
+      "loss": 0.0833,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022073170731707313,
+      "loss": 0.0337,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002219512195121951,
+      "loss": 0.0388,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022317073170731704,
+      "loss": 0.0646,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000224390243902439,
+      "loss": 0.0507,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022560975609756096,
+      "loss": 0.0264,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022682926829268292,
+      "loss": 0.0459,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022804878048780485,
+      "loss": 0.0394,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022926829268292683,
+      "loss": 0.0814,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023048780487804876,
+      "loss": 0.0205,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002317073170731707,
+      "loss": 0.0395,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023292682926829265,
+      "loss": 0.0781,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002341463414634146,
+      "loss": 0.0435,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023536585365853657,
+      "loss": 0.0715,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002365853658536585,
+      "loss": 0.0657,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023780487804878048,
+      "loss": 0.0321,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002390243902439024,
+      "loss": 0.0472,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024024390243902437,
+      "loss": 0.0425,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024146341463414633,
+      "loss": 0.0681,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024268292682926829,
+      "loss": 0.0902,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024390243902439022,
+      "loss": 0.054,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002451219512195122,
+      "loss": 0.0395,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024634146341463416,
+      "loss": 0.0823,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002475609756097561,
+      "loss": 0.0908,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000248780487804878,
+      "loss": 0.0745,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025,
+      "loss": 0.0415,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025121951219512194,
+      "loss": 0.0727,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025243902439024387,
+      "loss": 0.053,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025365853658536585,
+      "loss": 0.0292,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002548780487804878,
+      "loss": 0.0401,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002560975609756097,
+      "loss": 0.0338,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025731707317073164,
+      "loss": 0.0563,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025853658536585363,
+      "loss": 0.0173,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025975609756097556,
+      "loss": 0.0943,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026097560975609754,
+      "loss": 0.0493,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002621951219512195,
+      "loss": 0.0565,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026341463414634146,
+      "loss": 0.0507,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002646341463414634,
+      "loss": 0.0389,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002658536585365854,
+      "loss": 0.1022,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002670731707317073,
+      "loss": 0.0296,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026829268292682924,
+      "loss": 0.1012,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026951219512195117,
+      "loss": 0.0203,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027073170731707315,
+      "loss": 0.048,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002719512195121951,
+      "loss": 0.0479,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000273170731707317,
+      "loss": 0.0735,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000274390243902439,
+      "loss": 0.0365,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027560975609756093,
+      "loss": 0.0195,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002768292682926829,
+      "loss": 0.0254,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027804878048780484,
+      "loss": 0.0298,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027926829268292683,
+      "loss": 0.0459,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028048780487804876,
+      "loss": 0.0155,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002817073170731707,
+      "loss": 0.0237,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002829268292682927,
+      "loss": 0.0715,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002841463414634146,
+      "loss": 0.0456,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028536585365853654,
+      "loss": 0.077,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002865853658536585,
+      "loss": 0.0776,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028780487804878045,
+      "loss": 0.0948,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002890243902439024,
+      "loss": 0.1215,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029024390243902437,
+      "loss": 0.0517,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002914634146341463,
+      "loss": 0.0695,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002926829268292683,
+      "loss": 0.0677,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002939024390243902,
+      "loss": 0.0551,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002951219512195122,
+      "loss": 0.0429,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029634146341463413,
+      "loss": 0.0761,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029756097560975606,
+      "loss": 0.0387,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029878048780487804,
+      "loss": 0.0343,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003,
+      "loss": 0.0775,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999987473584,
+      "loss": 0.0449,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999999498943374,
+      "loss": 0.0312,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999887262259,
+      "loss": 0.0471,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999997995773525,
+      "loss": 0.066,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999686839617,
+      "loss": 0.0703,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999549049056,
+      "loss": 0.0565,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000299999938620567,
+      "loss": 0.0257,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999198309464,
+      "loss": 0.0553,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999998985360439,
+      "loss": 0.0673,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999987473586,
+      "loss": 0.0649,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000299999848430395,
+      "loss": 0.0242,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999981961964945,
+      "loss": 0.0248,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999997883036237,
+      "loss": 0.0472,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999997544823184,
+      "loss": 0.0406,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999971815573406,
+      "loss": 0.0873,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999996793238712,
+      "loss": 0.0659,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999996379867306,
+      "loss": 0.0276,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999995941443129,
+      "loss": 0.0365,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999954779661884,
+      "loss": 0.0623,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999994989436492,
+      "loss": 0.0676,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999994475854048,
+      "loss": 0.0449,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999993937218865,
+      "loss": 0.0699,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999933735309507,
+      "loss": 0.044,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999992784790316,
+      "loss": 0.0237,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999921709969706,
+      "loss": 0.1078,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999991532150925,
+      "loss": 0.0534,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999908682521885,
+      "loss": 0.0618,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999990179300774,
+      "loss": 0.0366,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999894652966915,
+      "loss": 0.0702,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999988726239953,
+      "loss": 0.0368,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999879621305723,
+      "loss": 0.0335,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999987172968561,
+      "loss": 0.0244,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999986358753932,
+      "loss": 0.0224,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999855194867,
+      "loss": 0.0277,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999846551668775,
+      "loss": 0.0556,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999837657944807,
+      "loss": 0.0783,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999982851369523,
+      "loss": 0.0527,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999819118920205,
+      "loss": 0.0449,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999809473619886,
+      "loss": 0.0479,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999979957779444,
+      "loss": 0.0295,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999978943144402,
+      "loss": 0.0519,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999977903456881,
+      "loss": 0.0255,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999768387168967,
+      "loss": 0.045,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999975748924468,
+      "loss": 0.0483,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999746340796134,
+      "loss": 0.0377,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999734941823505,
+      "loss": 0.0152,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999972329232699,
+      "loss": 0.0926,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999711392306785,
+      "loss": 0.0415,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999699241763084,
+      "loss": 0.0328,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999968684069609,
+      "loss": 0.0475,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999967418910601,
+      "loss": 0.0329,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999966128699306,
+      "loss": 0.0323,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999648134357456,
+      "loss": 0.0461,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999634731199407,
+      "loss": 0.0906,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999962107751915,
+      "loss": 0.0349,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299996071733169,
+      "loss": 0.0199,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999593018592905,
+      "loss": 0.0675,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999957861334739,
+      "loss": 0.0388,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299995639575806,
+      "loss": 0.0593,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999549051292775,
+      "loss": 0.0317,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999533894484165,
+      "loss": 0.0267,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999951848715503,
+      "loss": 0.0891,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999950282930562,
+      "loss": 0.0234,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299994869209362,
+      "loss": 0.0385,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999947076204704,
+      "loss": 0.0474,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299994543526384,
+      "loss": 0.0089,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999943769271056,
+      "loss": 0.0439,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999420782263794,
+      "loss": 0.0282,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999940362129839,
+      "loss": 0.0715,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999938620981463,
+      "loss": 0.0594,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999368547812806,
+      "loss": 0.0226,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999350635293216,
+      "loss": 0.0359,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999332472256164,
+      "loss": 0.0433,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999931405870194,
+      "loss": 0.0163,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999295394630857,
+      "loss": 0.1083,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999927648004323,
+      "loss": 0.0328,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999925731493937,
+      "loss": 0.0299,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000299992378993196,
+      "loss": 0.0094,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999921823318425,
+      "loss": 0.0228,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999919831653363,
+      "loss": 0.0682,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999178149368096,
+      "loss": 0.0442,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999157731687973,
+      "loss": 0.017,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000299991370634936,
+      "loss": 0.0479,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999911614478533,
+      "loss": 0.1235,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999094975563496,
+      "loss": 0.0237,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999907355582847,
+      "loss": 0.0284,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999905188558061,
+      "loss": 0.0566,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999902996482026,
+      "loss": 0.0413,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999007793547804,
+      "loss": 0.0412,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998985371763604,
+      "loss": 0.0257,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998962699468037,
+      "loss": 0.0218,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998939776661475,
+      "loss": 0.059,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999891660334431,
+      "loss": 0.0482,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999889317951692,
+      "loss": 0.0414,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999886950517971,
+      "loss": 0.0316,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999884558033306,
+      "loss": 0.0361,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999882140497738,
+      "loss": 0.1174,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999879697911307,
+      "loss": 0.0333,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998772302740535,
+      "loss": 0.0255,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999874737586019,
+      "loss": 0.0219,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998722198472453,
+      "loss": 0.0753,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998696770577746,
+      "loss": 0.0571,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999867109217649,
+      "loss": 0.0328,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999864516326912,
+      "loss": 0.0689,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998618983856053,
+      "loss": 0.0738,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999859255393774,
+      "loss": 0.0806,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998565873514625,
+      "loss": 0.0297,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998538942587144,
+      "loss": 0.0424,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999851176115575,
+      "loss": 0.0373,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000299984843292209,
+      "loss": 0.0167,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998456646783053,
+      "loss": 0.0345,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998428713842664,
+      "loss": 0.0279,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000299984005304002,
+      "loss": 0.041,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998372096456147,
+      "loss": 0.0473,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999834341201096,
+      "loss": 0.043,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999831447706513,
+      "loss": 0.0319,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999828529161913,
+      "loss": 0.1028,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999825585567347,
+      "loss": 0.0416,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999822616922861,
+      "loss": 0.0234,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999819623228507,
+      "loss": 0.0228,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999816604484334,
+      "loss": 0.0734,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999813560690392,
+      "loss": 0.0243,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999810491846733,
+      "loss": 0.0547,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029998073979534086,
+      "loss": 0.0342,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029998042790104683,
+      "loss": 0.0385,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999801135017966,
+      "loss": 0.0381,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999797965975953,
+      "loss": 0.0512,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997947718844834,
+      "loss": 0.0407,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299979155274361,
+      "loss": 0.0639,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999788308553387,
+      "loss": 0.0492,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999785039313868,
+      "loss": 0.0318,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997817450251076,
+      "loss": 0.0328,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997784256871607,
+      "loss": 0.0309,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999775081300084,
+      "loss": 0.0146,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999771711863932,
+      "loss": 0.0512,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997683173787603,
+      "loss": 0.0163,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999764897844628,
+      "loss": 0.065,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997614532615907,
+      "loss": 0.0522,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999757983629706,
+      "loss": 0.0182,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999754488949032,
+      "loss": 0.0309,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997509692196266,
+      "loss": 0.0864,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299974742444155,
+      "loss": 0.05,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999743854614859,
+      "loss": 0.0295,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999740259739616,
+      "loss": 0.0582,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999736639815879,
+      "loss": 0.0497,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299973299484371,
+      "loss": 0.0291,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999729324823168,
+      "loss": 0.0294,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997256297543154,
+      "loss": 0.0435,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999721909637214,
+      "loss": 0.0551,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997181644719263,
+      "loss": 0.0795,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999714394258514,
+      "loss": 0.0455,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997105989970407,
+      "loss": 0.066,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999706778687569,
+      "loss": 0.041,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997029333301634,
+      "loss": 0.0422,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999699062924888,
+      "loss": 0.0287,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996951674718077,
+      "loss": 0.0392,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996912469709865,
+      "loss": 0.0752,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999687301422491,
+      "loss": 0.0473,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996833308263863,
+      "loss": 0.0237,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996793351827396,
+      "loss": 0.024,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999675314491617,
+      "loss": 0.0357,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999671268753086,
+      "loss": 0.1922,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999667197967214,
+      "loss": 0.0469,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996631021340686,
+      "loss": 0.0371,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999658981253719,
+      "loss": 0.0494,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999654835326233,
+      "loss": 0.0267,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999650664351681,
+      "loss": 0.0336,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999646468330132,
+      "loss": 0.0438,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996422472616563,
+      "loss": 0.0468,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996380011463236,
+      "loss": 0.0346,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999633729984206,
+      "loss": 0.0458,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996294337753746,
+      "loss": 0.0544,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996251125199004,
+      "loss": 0.0285,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996207662178566,
+      "loss": 0.0359,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999616394869315,
+      "loss": 0.0586,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996119984743494,
+      "loss": 0.0186,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999607577033032,
+      "loss": 0.0878,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999603130545437,
+      "loss": 0.0458,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299959865901164,
+      "loss": 0.0275,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999594162431714,
+      "loss": 0.047,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995896408057353,
+      "loss": 0.039,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999585094133779,
+      "loss": 0.0205,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999580522415921,
+      "loss": 0.0461,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999575925652237,
+      "loss": 0.0402,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995713038428047,
+      "loss": 0.0374,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999566656987701,
+      "loss": 0.0364,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995619850870035,
+      "loss": 0.0323,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995572881407904,
+      "loss": 0.019,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995525661491397,
+      "loss": 0.0288,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995478191121303,
+      "loss": 0.0924,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995430470298426,
+      "loss": 0.0281,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995382499023546,
+      "loss": 0.0294,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999533427729747,
+      "loss": 0.0258,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995285805121014,
+      "loss": 0.0251,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995237082494977,
+      "loss": 0.0313,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995188109420174,
+      "loss": 0.0313,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999513888589743,
+      "loss": 0.1238,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999508941192755,
+      "loss": 0.0306,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999503968751138,
+      "loss": 0.0107,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999498971264974,
+      "loss": 0.0273,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994939487343466,
+      "loss": 0.0164,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299948890115934,
+      "loss": 0.0267,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999483828540038,
+      "loss": 0.0129,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999478730876526,
+      "loss": 0.0254,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999473608168888,
+      "loss": 0.031,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994684604172105,
+      "loss": 0.0982,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299946328762158,
+      "loss": 0.0381,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994580897820813,
+      "loss": 0.0204,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994528668988025,
+      "loss": 0.0847,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299944761897183,
+      "loss": 0.0379,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999442346001252,
+      "loss": 0.0463,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994370479871564,
+      "loss": 0.0167,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994317249296327,
+      "loss": 0.0333,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994263768287675,
+      "loss": 0.0304,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999421003684652,
+      "loss": 0.0495,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994156054973756,
+      "loss": 0.0307,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999410182267028,
+      "loss": 0.0239,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994047339936995,
+      "loss": 0.0465,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993992606774825,
+      "loss": 0.0422,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993937623184666,
+      "loss": 0.0216,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993882389167453,
+      "loss": 0.0112,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299938269047241,
+      "loss": 0.0537,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999377116985553,
+      "loss": 0.0138,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993715184562685,
+      "loss": 0.0543,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993658948846493,
+      "loss": 0.0188,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999360246270789,
+      "loss": 0.0416,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999354572614783,
+      "loss": 0.0357,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999348873916725,
+      "loss": 0.0678,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993431501767104,
+      "loss": 0.0173,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999337401394835,
+      "loss": 0.0153,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993316275711947,
+      "loss": 0.0297,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993258287058864,
+      "loss": 0.0197,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993200047990067,
+      "loss": 0.0749,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999314155850652,
+      "loss": 0.0162,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993082818609214,
+      "loss": 0.0192,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999302382829912,
+      "loss": 0.0219,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999296458757723,
+      "loss": 0.0655,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992905096444533,
+      "loss": 0.035,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999284535490201,
+      "loss": 0.0394,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999278536295068,
+      "loss": 0.0267,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992725120591524,
+      "loss": 0.1209,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999266462782556,
+      "loss": 0.0161,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299926038846538,
+      "loss": 0.052,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999254289107725,
+      "loss": 0.0173,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992481647096936,
+      "loss": 0.0156,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999242015271388,
+      "loss": 0.0196,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299923584079291,
+      "loss": 0.0817,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999229641274364,
+      "loss": 0.0555,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992234167158533,
+      "loss": 0.0125,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999217167117481,
+      "loss": 0.0328,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992108924793524,
+      "loss": 0.0158,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992045928015716,
+      "loss": 0.0369,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991982680842446,
+      "loss": 0.0415,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999191918327476,
+      "loss": 0.0427,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991855435313727,
+      "loss": 0.0343,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999179143696041,
+      "loss": 0.038,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991727188215874,
+      "loss": 0.0356,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299916626890812,
+      "loss": 0.0559,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999159793955746,
+      "loss": 0.0313,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999153293964573,
+      "loss": 0.042,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991467689347105,
+      "loss": 0.0371,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999140218866267,
+      "loss": 0.0455,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991336437593525,
+      "loss": 0.0172,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999127043614076,
+      "loss": 0.0225,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991204184305473,
+      "loss": 0.078,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999113768208879,
+      "loss": 0.0672,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991070929491807,
+      "loss": 0.0223,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991003926515636,
+      "loss": 0.0547,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999093667316141,
+      "loss": 0.0291,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999086916943024,
+      "loss": 0.0329,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990801415323256,
+      "loss": 0.0163,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990733410841593,
+      "loss": 0.0777,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990665155986385,
+      "loss": 0.0461,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999059665075877,
+      "loss": 0.0131,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990527895159896,
+      "loss": 0.0534,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999045888919091,
+      "loss": 0.0393,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990389632852965,
+      "loss": 0.0131,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990320126147213,
+      "loss": 0.0243,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990250369074824,
+      "loss": 0.0116,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990180361636956,
+      "loss": 0.0492,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999011010383478,
+      "loss": 0.041,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999003959566947,
+      "loss": 0.0407,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299899688371422,
+      "loss": 0.2534,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989897828254156,
+      "loss": 0.027,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989826569006527,
+      "loss": 0.0198,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989755059400494,
+      "loss": 0.058,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989683299437255,
+      "loss": 0.0579,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989611289118014,
+      "loss": 0.0291,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989539028443964,
+      "loss": 0.0258,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989466517416324,
+      "loss": 0.073,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299893937560363,
+      "loss": 0.0203,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299893207443051,
+      "loss": 0.0359,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998924748222395,
+      "loss": 0.0169,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989173969794075,
+      "loss": 0.1368,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989100207016695,
+      "loss": 0.1126,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989026193893054,
+      "loss": 0.0576,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998895193042438,
+      "loss": 0.0325,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998887741661192,
+      "loss": 0.0252,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029988802652456905,
+      "loss": 0.0448,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299887276379606,
+      "loss": 0.0238,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998865237312424,
+      "loss": 0.0453,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299885768579491,
+      "loss": 0.0385,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998850109243643,
+      "loss": 0.0537,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299884250765875,
+      "loss": 0.0779,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998834881040358,
+      "loss": 0.0126,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029988272293885945,
+      "loss": 0.0237,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998819552703587,
+      "loss": 0.0321,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998811850985463,
+      "loss": 0.0117,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998804124234353,
+      "loss": 0.019,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998796372450384,
+      "loss": 0.0246,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998788595633687,
+      "loss": 0.0217,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987807937843905,
+      "loss": 0.0308,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987729669026257,
+      "loss": 0.0623,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987651149885236,
+      "loss": 0.0433,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987572380422154,
+      "loss": 0.027,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987493360638317,
+      "loss": 0.0382,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998741409053505,
+      "loss": 0.0132,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987334570113675,
+      "loss": 0.0408,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029987254799375526,
+      "loss": 0.0428,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998717477832193,
+      "loss": 0.0393,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029987094506954225,
+      "loss": 0.0284,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998701398527375,
+      "loss": 0.0497,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998693321328185,
+      "loss": 0.0401,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998685219097988,
+      "loss": 0.0142,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986770918369185,
+      "loss": 0.0416,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998668939545113,
+      "loss": 0.061,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998660762222707,
+      "loss": 0.0388,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986525598698374,
+      "loss": 0.0257,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986443324866416,
+      "loss": 0.0392,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998636080073256,
+      "loss": 0.0168,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998627802629819,
+      "loss": 0.0471,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986195001564694,
+      "loss": 0.0265,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998611172653345,
+      "loss": 0.027,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998602820120585,
+      "loss": 0.0864,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998594442558329,
+      "loss": 0.0415,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998586039966717,
+      "loss": 0.0397,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299857761234589,
+      "loss": 0.0205,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998569159695987,
+      "loss": 0.1261,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998560682017151,
+      "loss": 0.0415,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985521793095233,
+      "loss": 0.0407,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985436515732447,
+      "loss": 0.0715,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985350988084587,
+      "loss": 0.0418,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998526521015308,
+      "loss": 0.0415,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985179181939354,
+      "loss": 0.0334,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998509290344485,
+      "loss": 0.0502,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998500637467101,
+      "loss": 0.0339,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984919595619277,
+      "loss": 0.0311,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299848325662911,
+      "loss": 0.0446,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998474528668793,
+      "loss": 0.0215,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984657756811236,
+      "loss": 0.1002,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984569976662466,
+      "loss": 0.0143,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998448194624309,
+      "loss": 0.0291,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984393665554587,
+      "loss": 0.0712,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998430513459842,
+      "loss": 0.0251,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984216353376073,
+      "loss": 0.0355,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998412732188903,
+      "loss": 0.0273,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984038040138775,
+      "loss": 0.0589,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299839485081268,
+      "loss": 0.0755,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299838587258546,
+      "loss": 0.0361,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998376869332368,
+      "loss": 0.0464,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983678410535535,
+      "loss": 0.0323,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983587877491675,
+      "loss": 0.0538,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998349709419362,
+      "loss": 0.0169,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983406060642875,
+      "loss": 0.0379,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983314776840965,
+      "loss": 0.0193,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998322324278942,
+      "loss": 0.0131,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998313145848976,
+      "loss": 0.0596,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998303942394352,
+      "loss": 0.0403,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982947139152243,
+      "loss": 0.0151,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982854604117465,
+      "loss": 0.0403,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998276181884074,
+      "loss": 0.0158,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000299826687833236,
+      "loss": 0.0276,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982575497567614,
+      "loss": 0.0393,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982481961574335,
+      "loss": 0.0806,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982388175345324,
+      "loss": 0.037,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982294138882154,
+      "loss": 0.0247,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982199852186385,
+      "loss": 0.0458,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000299821053152596,
+      "loss": 0.0196,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998201052810338,
+      "loss": 0.0942,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981915490719297,
+      "loss": 0.0228,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998182020310895,
+      "loss": 0.0321,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998172466527392,
+      "loss": 0.0358,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981628877215815,
+      "loss": 0.0452,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981532838936224,
+      "loss": 0.0433,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998143655043676,
+      "loss": 0.0464,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981340011719015,
+      "loss": 0.0436,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998124322278462,
+      "loss": 0.0535,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981146183635186,
+      "loss": 0.0148,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998104889427233,
+      "loss": 0.0248,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980951354697676,
+      "loss": 0.0437,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980853564912855,
+      "loss": 0.0598,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000299807555249195,
+      "loss": 0.0357,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998065723471926,
+      "loss": 0.0447,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980558694313753,
+      "loss": 0.0148,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980459903704645,
+      "loss": 0.0184,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998036086289358,
+      "loss": 0.0552,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998026157188221,
+      "loss": 0.0134,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980162030672193,
+      "loss": 0.0302,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998006223926519,
+      "loss": 0.0338,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997996219766288,
+      "loss": 0.032,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997986190586692,
+      "loss": 0.033,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997976136387899,
+      "loss": 0.0426,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979660571700765,
+      "loss": 0.0174,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979559529333936,
+      "loss": 0.0441,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979458236780186,
+      "loss": 0.0227,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979356694041206,
+      "loss": 0.0296,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000299792549011187,
+      "loss": 0.0278,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997915285801435,
+      "loss": 0.0283,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979050564729885,
+      "loss": 0.0601,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978948021266993,
+      "loss": 0.0247,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978845227627394,
+      "loss": 0.0216,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997874218381281,
+      "loss": 0.0345,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978638889824956,
+      "loss": 0.0316,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978535345665555,
+      "loss": 0.0124,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997843155133634,
+      "loss": 0.0101,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997832750683905,
+      "loss": 0.0197,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997822321217541,
+      "loss": 0.023,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029978118667347176,
+      "loss": 0.0145,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029978013872356083,
+      "loss": 0.0282,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997790882720388,
+      "loss": 0.0491,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977803531892335,
+      "loss": 0.0308,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977697986423193,
+      "loss": 0.0098,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977592190798225,
+      "loss": 0.0704,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000299774861450192,
+      "loss": 0.0212,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977379849087875,
+      "loss": 0.0185,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997727330300604,
+      "loss": 0.0257,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977166506775467,
+      "loss": 0.0259,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977059460397943,
+      "loss": 0.018,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029976952163875256,
+      "loss": 0.0249,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029976844617209196,
+      "loss": 0.0179,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997673682040156,
+      "loss": 0.0141,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997662877345415,
+      "loss": 0.0367,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997652047636877,
+      "loss": 0.0335,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997641192914722,
+      "loss": 0.0296,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997630313179133,
+      "loss": 0.0328,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029976194084302903,
+      "loss": 0.0206,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029976084786683766,
+      "loss": 0.0216,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975975238935744,
+      "loss": 0.0287,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975865441060666,
+      "loss": 0.0182,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975755393060364,
+      "loss": 0.03,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997564509493668,
+      "loss": 0.0349,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997553454669146,
+      "loss": 0.043,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975423748326537,
+      "loss": 0.0541,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975312699843774,
+      "loss": 0.0298,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997520140124502,
+      "loss": 0.0168,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975089852532136,
+      "loss": 0.0332,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997497805370698,
+      "loss": 0.027,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029974866004771426,
+      "loss": 0.0309,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997475370572734,
+      "loss": 0.0244,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029974641156576605,
+      "loss": 0.0195,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997452835732109,
+      "loss": 0.0339,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997441530796269,
+      "loss": 0.0589,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029974302008503286,
+      "loss": 0.012,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997418845894477,
+      "loss": 0.0427,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029974074659289043,
+      "loss": 0.0414,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997396060953801,
+      "loss": 0.0322,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997384630969356,
+      "loss": 0.01,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997373175975762,
+      "loss": 0.0308,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997361695973209,
+      "loss": 0.0362,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997350190961889,
+      "loss": 0.0655,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973386609419945,
+      "loss": 0.0695,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997327105913718,
+      "loss": 0.0388,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997315525877253,
+      "loss": 0.028,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997303920832791,
+      "loss": 0.0412,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029972922907805285,
+      "loss": 0.0295,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029972806357206577,
+      "loss": 0.0816,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997268955653374,
+      "loss": 0.0325,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029972572505788725,
+      "loss": 0.0783,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997245520497349,
+      "loss": 0.0367,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029972337654089986,
+      "loss": 0.0079,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997221985314018,
+      "loss": 0.0603,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029972101802126045,
+      "loss": 0.0231,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997198350104954,
+      "loss": 0.0414,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997186494991266,
+      "loss": 0.0447,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997174614871737,
+      "loss": 0.0369,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971627097465656,
+      "loss": 0.0504,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971507796159506,
+      "loss": 0.0127,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997138824480092,
+      "loss": 0.0367,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997126844339189,
+      "loss": 0.0342,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971148391934413,
+      "loss": 0.0391,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971028090430504,
+      "loss": 0.0309,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997090753888216,
+      "loss": 0.0405,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000299707867372914,
+      "loss": 0.0288,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997066568566025,
+      "loss": 0.011,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970544383990715,
+      "loss": 0.0371,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970422832284834,
+      "loss": 0.024,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970301030544634,
+      "loss": 0.037,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997017897877214,
+      "loss": 0.0245,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970056676969413,
+      "loss": 0.0374,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969934125138474,
+      "loss": 0.0475,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969811323281374,
+      "loss": 0.0259,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969688271400173,
+      "loss": 0.0134,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969564969496915,
+      "loss": 0.0164,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969441417573666,
+      "loss": 0.0053,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969317615632495,
+      "loss": 0.0222,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996919356367546,
+      "loss": 0.0462,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996906926170463,
+      "loss": 0.0293,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029968944709722096,
+      "loss": 0.0102,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029968819907729924,
+      "loss": 0.0418,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996869485573021,
+      "loss": 0.0567,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996856955372503,
+      "loss": 0.0114,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996844400171649,
+      "loss": 0.0108,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996831819970667,
+      "loss": 0.0354,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029968192147697686,
+      "loss": 0.0417,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996806584569164,
+      "loss": 0.051,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996793929369064,
+      "loss": 0.0295,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967812491696805,
+      "loss": 0.0255,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967685439712236,
+      "loss": 0.026,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967558137739073,
+      "loss": 0.0236,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996743058577944,
+      "loss": 0.0405,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996730278383546,
+      "loss": 0.03,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996717473190927,
+      "loss": 0.1239,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967046430003005,
+      "loss": 0.0206,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029966917878118817,
+      "loss": 0.0353,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029966789076258846,
+      "loss": 0.0369,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996666002442525,
+      "loss": 0.0174,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996653072262018,
+      "loss": 0.0366,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029966401170845793,
+      "loss": 0.064,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029966271369104254,
+      "loss": 0.0226,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996614131739773,
+      "loss": 0.0349,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029966011015728405,
+      "loss": 0.0084,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996588046409844,
+      "loss": 0.0565,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029965749662510014,
+      "loss": 0.0095,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029965618610965333,
+      "loss": 0.0826,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996548730946656,
+      "loss": 0.0636,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029965355758015906,
+      "loss": 0.0113,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996522395661556,
+      "loss": 0.0346,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996509190526772,
+      "loss": 0.0271,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996495960397461,
+      "loss": 0.0381,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964827052738415,
+      "loss": 0.0404,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996469425156136,
+      "loss": 0.0437,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996456120044567,
+      "loss": 0.0462,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964427899393554,
+      "loss": 0.0222,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996429434840725,
+      "loss": 0.0427,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996416054748898,
+      "loss": 0.0245,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996402649664098,
+      "loss": 0.0216,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996389219586549,
+      "loss": 0.03,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996375764516476,
+      "loss": 0.0557,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963622844541026,
+      "loss": 0.0772,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996348779399655,
+      "loss": 0.0487,
+      "step": 786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963352493533576,
+      "loss": 0.0473,
+      "step": 787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963216943154375,
+      "loss": 0.0469,
+      "step": 788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996308114286121,
+      "loss": 0.0508,
+      "step": 789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962945092656333,
+      "loss": 0.039,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962808792542033,
+      "loss": 0.0405,
+      "step": 791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962672242520586,
+      "loss": 0.0416,
+      "step": 792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996253544259426,
+      "loss": 0.0296,
+      "step": 793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996239839276536,
+      "loss": 0.016,
+      "step": 794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962261093036157,
+      "loss": 0.0204,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996212354340895,
+      "loss": 0.0325,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996198574388604,
+      "loss": 0.0753,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961847694469725,
+      "loss": 0.0461,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996170939516231,
+      "loss": 0.0342,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961570845966104,
+      "loss": 0.0445,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961432046883427,
+      "loss": 0.0296,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996129299791658,
+      "loss": 0.0125,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961153699067917,
+      "loss": 0.0372,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961014150339734,
+      "loss": 0.0288,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996087435173438,
+      "loss": 0.0355,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996073430325418,
+      "loss": 0.009,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996059400490148,
+      "loss": 0.0416,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029960453456678615,
+      "loss": 0.0325,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029960312658587943,
+      "loss": 0.0363,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029960171610631805,
+      "loss": 0.0403,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029960030312812565,
+      "loss": 0.0262,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995988876513258,
+      "loss": 0.037,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959746967594215,
+      "loss": 0.0252,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959604920199833,
+      "loss": 0.016,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995946262295181,
+      "loss": 0.0371,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995932007585252,
+      "loss": 0.0137,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959177278904357,
+      "loss": 0.0085,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959034232109695,
+      "loss": 0.0232,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995889093547092,
+      "loss": 0.0628,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995874738899043,
+      "loss": 0.0109,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995860359267062,
+      "loss": 0.0215,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000299584595465139,
+      "loss": 0.0364,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995831525052266,
+      "loss": 0.0241,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029958170704699326,
+      "loss": 0.0532,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995802590904631,
+      "loss": 0.0646,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957880863566017,
+      "loss": 0.0351,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957735568260884,
+      "loss": 0.0538,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995759002313333,
+      "loss": 0.0193,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957444228185783,
+      "loss": 0.056,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957298183420694,
+      "loss": 0.0151,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995715188884049,
+      "loss": 0.0342,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995700534444761,
+      "loss": 0.0338,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995685855024451,
+      "loss": 0.0215,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995671150623364,
+      "loss": 0.0162,
+      "step": 834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995656421241745,
+      "loss": 0.0217,
+      "step": 835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029956416668798405,
+      "loss": 0.0401,
+      "step": 836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029956268875378975,
+      "loss": 0.0738,
+      "step": 837
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995612083216162,
+      "loss": 0.0364,
+      "step": 838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995597253914882,
+      "loss": 0.0432,
+      "step": 839
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995582399634304,
+      "loss": 0.04,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029955675203746776,
+      "loss": 0.1019,
+      "step": 841
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000299555261613625,
+      "loss": 0.0194,
+      "step": 842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995537686919271,
+      "loss": 0.0304,
+      "step": 843
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995522732723989,
+      "loss": 0.0208,
+      "step": 844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029955077535506546,
+      "loss": 0.0324,
+      "step": 845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995492749399518,
+      "loss": 0.0547,
+      "step": 846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029954777202708294,
+      "loss": 0.0198,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000299546266616484,
+      "loss": 0.0215,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995447587081801,
+      "loss": 0.0268,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995432483021965,
+      "loss": 0.0152,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995417353985583,
+      "loss": 0.0227,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995402199972909,
+      "loss": 0.0546,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953870209841946,
+      "loss": 0.0295,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995371817019695,
+      "loss": 0.0514,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995356588079663,
+      "loss": 0.0412,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953413341643535,
+      "loss": 0.0225,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995326055274021,
+      "loss": 0.0123,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953107514089206,
+      "loss": 0.013,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029952954225693086,
+      "loss": 0.011,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000299528006875544,
+      "loss": 0.0108,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029952646899675717,
+      "loss": 0.0363,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029952492862059607,
+      "loss": 0.0217,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995233857470864,
+      "loss": 0.0409,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029952184037625396,
+      "loss": 0.0524,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995202925081245,
+      "loss": 0.0636,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951874214272393,
+      "loss": 0.0259,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951718928007817,
+      "loss": 0.0419,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995156339202131,
+      "loss": 0.0169,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995140760631547,
+      "loss": 0.0421,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951251570892905,
+      "loss": 0.0188,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951095285756207,
+      "loss": 0.027,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950938750908005,
+      "loss": 0.0159,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950781966350896,
+      "loss": 0.0328,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950624932087515,
+      "loss": 0.048,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995046764812047,
+      "loss": 0.0166,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995031011445239,
+      "loss": 0.0297,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995015233108592,
+      "loss": 0.0411,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994999429802368,
+      "loss": 0.0398,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994983601526832,
+      "loss": 0.028,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029949677482822475,
+      "loss": 0.0097,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000299495187006888,
+      "loss": 0.1199,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994935966886994,
+      "loss": 0.0205,
+      "step": 882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029949200387368555,
+      "loss": 0.0107,
+      "step": 883
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029949040856187305,
+      "loss": 0.0171,
+      "step": 884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994888107532885,
+      "loss": 0.0431,
+      "step": 885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029948721044795873,
+      "loss": 0.0396,
+      "step": 886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994856076459103,
+      "loss": 0.0143,
+      "step": 887
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994840023471701,
+      "loss": 0.0127,
+      "step": 888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994823945517648,
+      "loss": 0.1093,
+      "step": 889
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029948078425972137,
+      "loss": 0.0169,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029947917147106674,
+      "loss": 0.0382,
+      "step": 891
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029947755618582773,
+      "loss": 0.0352,
+      "step": 892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994759384040314,
+      "loss": 0.0134,
+      "step": 893
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029947431812570475,
+      "loss": 0.0241,
+      "step": 894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994726953508748,
+      "loss": 0.0276,
+      "step": 895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994710700795687,
+      "loss": 0.0196,
+      "step": 896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994694423118136,
+      "loss": 0.0149,
+      "step": 897
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029946781204763664,
+      "loss": 0.0427,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029946617928706514,
+      "loss": 0.0332,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994645440301262,
+      "loss": 0.0441,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029946290627684736,
+      "loss": 0.0209,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994612660272558,
+      "loss": 0.0303,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945962328137895,
+      "loss": 0.0243,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945797803924425,
+      "loss": 0.0141,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945633030087923,
+      "loss": 0.0187,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945468006631137,
+      "loss": 0.0146,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945302733556825,
+      "loss": 0.0421,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994513721086775,
+      "loss": 0.0234,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029944971438566664,
+      "loss": 0.0551,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994480541665635,
+      "loss": 0.0537,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029944639145139575,
+      "loss": 0.0314,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029944472624019114,
+      "loss": 0.0122,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994430585329775,
+      "loss": 0.0172,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994413883297827,
+      "loss": 0.0479,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029943971563063464,
+      "loss": 0.0876,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994380404355612,
+      "loss": 0.0365,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994363627445904,
+      "loss": 0.032,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029943468255775034,
+      "loss": 0.017,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029943299987506895,
+      "loss": 0.0098,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029943131469657437,
+      "loss": 0.0201,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994296270222948,
+      "loss": 0.071,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994279368522584,
+      "loss": 0.0509,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942624418649333,
+      "loss": 0.013,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299424549025028,
+      "loss": 0.0351,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942285136789055,
+      "loss": 0.0187,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942115121510946,
+      "loss": 0.0498,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994194485667131,
+      "loss": 0.0208,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994177434227299,
+      "loss": 0.0641,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029941603578318836,
+      "loss": 0.0428,
+      "step": 929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299414325648117,
+      "loss": 0.0225,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994126130175443,
+      "loss": 0.0417,
+      "step": 931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994108978914989,
+      "loss": 0.0198,
+      "step": 932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994091802700096,
+      "loss": 0.0406,
+      "step": 933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994074601531048,
+      "loss": 0.0073,
+      "step": 934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994057375408135,
+      "loss": 0.0173,
+      "step": 935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994040124331644,
+      "loss": 0.0643,
+      "step": 936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994022848301862,
+      "loss": 0.0251,
+      "step": 937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994005547319078,
+      "loss": 0.0148,
+      "step": 938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029939882213835813,
+      "loss": 0.0332,
+      "step": 939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029939708704956614,
+      "loss": 0.044,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993953494655608,
+      "loss": 0.0091,
+      "step": 941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993936093863711,
+      "loss": 0.031,
+      "step": 942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029939186681202616,
+      "loss": 0.1274,
+      "step": 943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299390121742555,
+      "loss": 0.0566,
+      "step": 944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993883741779868,
+      "loss": 0.0291,
+      "step": 945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993866241183508,
+      "loss": 0.0287,
+      "step": 946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029938487156367624,
+      "loss": 0.0353,
+      "step": 947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029938311651399226,
+      "loss": 0.0152,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993813589693283,
+      "loss": 0.0099,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029937959892971367,
+      "loss": 0.0122,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993778363951777,
+      "loss": 0.022,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993760713657499,
+      "loss": 0.0586,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993743038414598,
+      "loss": 0.0309,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029937253382233687,
+      "loss": 0.0688,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993707613084106,
+      "loss": 0.0254,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993689862997107,
+      "loss": 0.0831,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029936720879626677,
+      "loss": 0.0339,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993654287981085,
+      "loss": 0.0133,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993636463052656,
+      "loss": 0.0186,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993618613177679,
+      "loss": 0.0345,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029936007383564515,
+      "loss": 0.0288,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029935828385892725,
+      "loss": 0.0329,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029935649138764403,
+      "loss": 0.0164,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029935469642182553,
+      "loss": 0.0177,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993528989615016,
+      "loss": 0.018,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993510990067024,
+      "loss": 0.0508,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993492965574579,
+      "loss": 0.0167,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993474916137983,
+      "loss": 0.0174,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993456841757536,
+      "loss": 0.0286,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029934387424335407,
+      "loss": 0.0337,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029934206181662996,
+      "loss": 0.0391,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993402468956115,
+      "loss": 0.0324,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000299338429480329,
+      "loss": 0.0938,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029933660957081286,
+      "loss": 0.0297,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029933478716709345,
+      "loss": 0.0291,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993329622692012,
+      "loss": 0.0125,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993311348771666,
+      "loss": 0.0538,
+      "step": 977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993293049910201,
+      "loss": 0.0194,
+      "step": 978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993274726107924,
+      "loss": 0.0325,
+      "step": 979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029932563773651396,
+      "loss": 0.0112,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993238003682156,
+      "loss": 0.0143,
+      "step": 981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993219605059278,
+      "loss": 0.0374,
+      "step": 982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993201181496815,
+      "loss": 0.0258,
+      "step": 983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029931827329950723,
+      "loss": 0.0235,
+      "step": 984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029931642595543606,
+      "loss": 0.029,
+      "step": 985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029931457611749867,
+      "loss": 0.0441,
+      "step": 986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000299312723785726,
+      "loss": 0.0173,
+      "step": 987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029931086896014907,
+      "loss": 0.0475,
+      "step": 988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993090116407987,
+      "loss": 0.0489,
+      "step": 989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000299307151827706,
+      "loss": 0.0337,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993052895209021,
+      "loss": 0.0263,
+      "step": 991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029930342472041796,
+      "loss": 0.0647,
+      "step": 992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993015574262848,
+      "loss": 0.0486,
+      "step": 993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029929968763853384,
+      "loss": 0.0253,
+      "step": 994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992978153571963,
+      "loss": 0.0089,
+      "step": 995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029929594058230333,
+      "loss": 0.0169,
+      "step": 996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992940633138864,
+      "loss": 0.011,
+      "step": 997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992921835519767,
+      "loss": 0.0192,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992903012966059,
+      "loss": 0.0343,
+      "step": 999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029928841654780513,
+      "loss": 0.0999,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992865293056061,
+      "loss": 0.0121,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029928463957004013,
+      "loss": 0.0408,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029928274734113895,
+      "loss": 0.0253,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992808526189341,
+      "loss": 0.0261,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992789554034572,
+      "loss": 0.0291,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029927705569473997,
+      "loss": 0.0601,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992751534928141,
+      "loss": 0.0287,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029927324879771146,
+      "loss": 0.0151,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992713416094637,
+      "loss": 0.0269,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992694319281028,
+      "loss": 0.02,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029926751975366066,
+      "loss": 0.047,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992656050861691,
+      "loss": 0.0362,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029926368792566025,
+      "loss": 0.0204,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029926176827216604,
+      "loss": 0.0788,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925984612571855,
+      "loss": 0.0261,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992579214863499,
+      "loss": 0.0421,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925599435409213,
+      "loss": 0.022,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992540647289776,
+      "loss": 0.028,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925213261103843,
+      "loss": 0.0803,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925019800030694,
+      "loss": 0.0408,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992482608968154,
+      "loss": 0.0088,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029924632130059613,
+      "loss": 0.0283,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992443792116816,
+      "loss": 0.0096,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992424346301042,
+      "loss": 0.0151,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029924048755589647,
+      "loss": 0.0586,
+      "step": 1025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029923853798909084,
+      "loss": 0.0282,
+      "step": 1026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992365859297199,
+      "loss": 0.0243,
+      "step": 1027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992346313778164,
+      "loss": 0.0313,
+      "step": 1028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992326743334127,
+      "loss": 0.0223,
+      "step": 1029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992307147965418,
+      "loss": 0.033,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922875276723615,
+      "loss": 0.0164,
+      "step": 1031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992267882455286,
+      "loss": 0.0289,
+      "step": 1032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992248212314521,
+      "loss": 0.0206,
+      "step": 1033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992228517250394,
+      "loss": 0.0388,
+      "step": 1034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922087972632336,
+      "loss": 0.0074,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921890523533696,
+      "loss": 0.0461,
+      "step": 1036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921692825211313,
+      "loss": 0.0137,
+      "step": 1037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000299214948776685,
+      "loss": 0.0109,
+      "step": 1038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992129668090855,
+      "loss": 0.0362,
+      "step": 1039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921098234934787,
+      "loss": 0.0648,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992089953975051,
+      "loss": 0.0326,
+      "step": 1041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029920700595359055,
+      "loss": 0.0308,
+      "step": 1042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992050140176373,
+      "loss": 0.0315,
+      "step": 1043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992030195896787,
+      "loss": 0.0216,
+      "step": 1044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000299201022669748,
+      "loss": 0.0097,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991990232578786,
+      "loss": 0.0193,
+      "step": 1046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991970213541039,
+      "loss": 0.082,
+      "step": 1047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029919501695845725,
+      "loss": 0.0219,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991930100709722,
+      "loss": 0.0216,
+      "step": 1049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029919100069168226,
+      "loss": 0.0394,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000299188988820621,
+      "loss": 0.0525,
+      "step": 1051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029918697445782206,
+      "loss": 0.0052,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000299184957603319,
+      "loss": 0.0644,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029918293825714553,
+      "loss": 0.05,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991809164193354,
+      "loss": 0.0306,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991788920899224,
+      "loss": 0.0426,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991768652689403,
+      "loss": 0.0056,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029917483595642293,
+      "loss": 0.0134,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991728041524042,
+      "loss": 0.0369,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991707698569181,
+      "loss": 0.0306,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029916873306999847,
+      "loss": 0.009,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029916669379167956,
+      "loss": 0.0147,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991646520219952,
+      "loss": 0.0318,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029916260776097957,
+      "loss": 0.0146,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991605610086669,
+      "loss": 0.033,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029915851176509124,
+      "loss": 0.0121,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991564600302869,
+      "loss": 0.0536,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991544058042881,
+      "loss": 0.018,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991523490871292,
+      "loss": 0.0421,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029915028987884454,
+      "loss": 0.0165,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991482281794685,
+      "loss": 0.0337,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991461639890355,
+      "loss": 0.0248,
+      "step": 1072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029914409730758,
+      "loss": 0.0178,
+      "step": 1073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029914202813513654,
+      "loss": 0.034,
+      "step": 1074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991399564717397,
+      "loss": 0.0216,
+      "step": 1075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991378823174241,
+      "loss": 0.0225,
+      "step": 1076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991358056722243,
+      "loss": 0.017,
+      "step": 1077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029913372653617506,
+      "loss": 0.0257,
+      "step": 1078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991316449093111,
+      "loss": 0.0131,
+      "step": 1079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029912956079166715,
+      "loss": 0.0362,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000299127474183278,
+      "loss": 0.0309,
+      "step": 1081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029912538508417854,
+      "loss": 0.0258,
+      "step": 1082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991232934944037,
+      "loss": 0.0073,
+      "step": 1083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991211994139883,
+      "loss": 0.0065,
+      "step": 1084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029911910284296747,
+      "loss": 0.032,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029911700378137604,
+      "loss": 0.0233,
+      "step": 1086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991149022292492,
+      "loss": 0.0878,
+      "step": 1087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991127981866221,
+      "loss": 0.017,
+      "step": 1088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991106916535297,
+      "loss": 0.0169,
+      "step": 1089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991085826300073,
+      "loss": 0.0286,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910647111609014,
+      "loss": 0.0355,
+      "step": 1091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910435711181344,
+      "loss": 0.0286,
+      "step": 1092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991022406172125,
+      "loss": 0.0353,
+      "step": 1093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910012163232276,
+      "loss": 0.0502,
+      "step": 1094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990980001571795,
+      "loss": 0.1432,
+      "step": 1095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990958761918182,
+      "loss": 0.0148,
+      "step": 1096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990937497362743,
+      "loss": 0.0159,
+      "step": 1097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029909162079058335,
+      "loss": 0.057,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990894893547809,
+      "loss": 0.0491,
+      "step": 1099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990873554289026,
+      "loss": 0.0141,
+      "step": 1100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000299085219012984,
+      "loss": 0.0504,
+      "step": 1101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990830801070608,
+      "loss": 0.0285,
+      "step": 1102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990809387111688,
+      "loss": 0.064,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990787948253437,
+      "loss": 0.0232,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990766484496213,
+      "loss": 0.0147,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990744995840375,
+      "loss": 0.038,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029907234822862817,
+      "loss": 0.0681,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990701943834292,
+      "loss": 0.0126,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990680380484766,
+      "loss": 0.0258,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990658792238064,
+      "loss": 0.0315,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029906371790945456,
+      "loss": 0.0767,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029906155410545736,
+      "loss": 0.0196,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990593878118508,
+      "loss": 0.0291,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029905721902867105,
+      "loss": 0.0392,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029905504775595436,
+      "loss": 0.0445,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990528739937371,
+      "loss": 0.0255,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990506977420554,
+      "loss": 0.0489,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029904851900094574,
+      "loss": 0.0429,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029904633777044445,
+      "loss": 0.0471,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000299044154050588,
+      "loss": 0.0388,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990419678414128,
+      "loss": 0.0194,
+      "step": 1121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990397791429554,
+      "loss": 0.0315,
+      "step": 1122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990375879552524,
+      "loss": 0.0514,
+      "step": 1123
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029903539427834035,
+      "loss": 0.0403,
+      "step": 1124
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029903319811225586,
+      "loss": 0.0561,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990309994570356,
+      "loss": 0.0283,
+      "step": 1126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990287983127164,
+      "loss": 0.0368,
+      "step": 1127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029902659467933494,
+      "loss": 0.0302,
+      "step": 1128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029902438855692806,
+      "loss": 0.0312,
+      "step": 1129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029902217994553257,
+      "loss": 0.0248,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029901996884518537,
+      "loss": 0.0465,
+      "step": 1131
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990177552559234,
+      "loss": 0.0238,
+      "step": 1132
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990155391777836,
+      "loss": 0.0675,
+      "step": 1133
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029901332061080303,
+      "loss": 0.0812,
+      "step": 1134
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990110995550187,
+      "loss": 0.0157,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029900887601046775,
+      "loss": 0.0493,
+      "step": 1136
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029900664997718735,
+      "loss": 0.0362,
+      "step": 1137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029900442145521454,
+      "loss": 0.0196,
+      "step": 1138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990021904445867,
+      "loss": 0.0473,
+      "step": 1139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000298999956945341,
+      "loss": 0.0122,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029899772095751473,
+      "loss": 0.0158,
+      "step": 1141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989954824811453,
+      "loss": 0.0353,
+      "step": 1142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029899324151627004,
+      "loss": 0.0873,
+      "step": 1143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029899099806292637,
+      "loss": 0.03,
+      "step": 1144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029898875212115185,
+      "loss": 0.0098,
+      "step": 1145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989865036909839,
+      "loss": 0.0391,
+      "step": 1146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989842527724601,
+      "loss": 0.0041,
+      "step": 1147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029898199936561807,
+      "loss": 0.0446,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029897974347049545,
+      "loss": 0.0178,
+      "step": 1149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989774850871299,
+      "loss": 0.0263,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029897522421555903,
+      "loss": 0.0635,
+      "step": 1151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989729608558208,
+      "loss": 0.0572,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989706950079529,
+      "loss": 0.0557,
+      "step": 1153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989684266719932,
+      "loss": 0.0415,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029896615584797955,
+      "loss": 0.0372,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029896388253594994,
+      "loss": 0.029,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029896160673594224,
+      "loss": 0.05,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029895932844799456,
+      "loss": 0.0549,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989570476721449,
+      "loss": 0.0215,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989547644084314,
+      "loss": 0.0485,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989524786568922,
+      "loss": 0.0341,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989501904175653,
+      "loss": 0.0294,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989478996904892,
+      "loss": 0.0158,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029894560647570197,
+      "loss": 0.0059,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989433107732419,
+      "loss": 0.0169,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029894101258314747,
+      "loss": 0.0151,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029893871190545695,
+      "loss": 0.0146,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989364087402088,
+      "loss": 0.02,
+      "step": 1168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989341030874415,
+      "loss": 0.0292,
+      "step": 1169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989317949471935,
+      "loss": 0.031,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029892948431950347,
+      "loss": 0.0299,
+      "step": 1171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029892717120440985,
+      "loss": 0.0247,
+      "step": 1172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029892485560195143,
+      "loss": 0.0119,
+      "step": 1173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989225375121667,
+      "loss": 0.0218,
+      "step": 1174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989202169350946,
+      "loss": 0.0194,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989178938707737,
+      "loss": 0.0545,
+      "step": 1176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029891556831924287,
+      "loss": 0.0085,
+      "step": 1177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000298913240280541,
+      "loss": 0.0224,
+      "step": 1178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029891090975470684,
+      "loss": 0.0168,
+      "step": 1179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890857674177946,
+      "loss": 0.0722,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890624124179774,
+      "loss": 0.0534,
+      "step": 1181
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890390325480076,
+      "loss": 0.079,
+      "step": 1182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890156278082744,
+      "loss": 0.0264,
+      "step": 1183
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000298899219819917,
+      "loss": 0.0852,
+      "step": 1184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988968743721085,
+      "loss": 0.0234,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988945264374412,
+      "loss": 0.0066,
+      "step": 1186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029889217601595413,
+      "loss": 0.0399,
+      "step": 1187
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888982310768676,
+      "loss": 0.0273,
+      "step": 1188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988874677126783,
+      "loss": 0.0274,
+      "step": 1189
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888510983096803,
+      "loss": 0.017,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988827494625954,
+      "loss": 0.0381,
+      "step": 1191
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888038660759984,
+      "loss": 0.0313,
+      "step": 1192
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029887802126602077,
+      "loss": 0.045,
+      "step": 1193
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029887565343789774,
+      "loss": 0.0209,
+      "step": 1194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988732831232702,
+      "loss": 0.0229,
+      "step": 1195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988709103221779,
+      "loss": 0.028,
+      "step": 1196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988685350346604,
+      "loss": 0.032,
+      "step": 1197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029886615726075727,
+      "loss": 0.0287,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029886377700050833,
+      "loss": 0.0305,
+      "step": 1199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029886139425395336,
+      "loss": 0.0106,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029885900902113205,
+      "loss": 0.019,
+      "step": 1201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988566213020844,
+      "loss": 0.0308,
+      "step": 1202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029885423109685013,
+      "loss": 0.0191,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029885183840546917,
+      "loss": 0.043,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029884944322798156,
+      "loss": 0.0501,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988470455644273,
+      "loss": 0.0363,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988446454148464,
+      "loss": 0.0266,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029884224277927895,
+      "loss": 0.0173,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988398376577651,
+      "loss": 0.034,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000298837430050345,
+      "loss": 0.017,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029883501995705885,
+      "loss": 0.0134,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029883260737794694,
+      "loss": 0.0387,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029883019231304953,
+      "loss": 0.0113,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000298827774762407,
+      "loss": 0.0278,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988253547260597,
+      "loss": 0.0144,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000298822932204048,
+      "loss": 0.0218,
+      "step": 1216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029882050719641246,
+      "loss": 0.0184,
+      "step": 1217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988180797031935,
+      "loss": 0.0199,
+      "step": 1218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029881564972443166,
+      "loss": 0.0178,
+      "step": 1219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029881321726016763,
+      "loss": 0.0557,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029881078231044195,
+      "loss": 0.0244,
+      "step": 1221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988083448752953,
+      "loss": 0.0134,
+      "step": 1222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029880590495476837,
+      "loss": 0.0162,
+      "step": 1223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029880346254890195,
+      "loss": 0.0121,
+      "step": 1224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029880101765773683,
+      "loss": 0.0298,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029879857028131384,
+      "loss": 0.0253,
+      "step": 1226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029879612041967384,
+      "loss": 0.0147,
+      "step": 1227
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987936680728578,
+      "loss": 0.0501,
+      "step": 1228
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987912132409065,
+      "loss": 0.0492,
+      "step": 1229
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987887559238612,
+      "loss": 0.0232,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987862961217628,
+      "loss": 0.0337,
+      "step": 1231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029878383383465237,
+      "loss": 0.0117,
+      "step": 1232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987813690625711,
+      "loss": 0.0092,
+      "step": 1233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029877890180556006,
+      "loss": 0.0296,
+      "step": 1234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987764320636606,
+      "loss": 0.046,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987739598369138,
+      "loss": 0.0241,
+      "step": 1236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029877148512536114,
+      "loss": 0.0483,
+      "step": 1237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987690079290438,
+      "loss": 0.0246,
+      "step": 1238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987665282480032,
+      "loss": 0.0164,
+      "step": 1239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029876404608228076,
+      "loss": 0.0345,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987615614319179,
+      "loss": 0.0329,
+      "step": 1241
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029875907429695623,
+      "loss": 0.0591,
+      "step": 1242
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987565846774372,
+      "loss": 0.012,
+      "step": 1243
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029875409257340243,
+      "loss": 0.048,
+      "step": 1244
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029875159798489347,
+      "loss": 0.0228,
+      "step": 1245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987491009119521,
+      "loss": 0.0139,
+      "step": 1246
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029874660135461987,
+      "loss": 0.0676,
+      "step": 1247
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029874409931293866,
+      "loss": 0.0142,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987415947869502,
+      "loss": 0.0096,
+      "step": 1249
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987390877766964,
+      "loss": 0.0451,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000298736578282219,
+      "loss": 0.0386,
+      "step": 1251
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029873406630356007,
+      "loss": 0.0276,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987315518407614,
+      "loss": 0.0215,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029872903489386505,
+      "loss": 0.0191,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987265154629131,
+      "loss": 0.0458,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987239935479476,
+      "loss": 0.0141,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987214691490107,
+      "loss": 0.088,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029871894226614446,
+      "loss": 0.0188,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029871641289939125,
+      "loss": 0.0594,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987138810487932,
+      "loss": 0.0223,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987113467143926,
+      "loss": 0.0579,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987088098962318,
+      "loss": 0.0116,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029870627059435317,
+      "loss": 0.0292,
+      "step": 1263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029870372880879905,
+      "loss": 0.0491,
+      "step": 1264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029870118453961206,
+      "loss": 0.0272,
+      "step": 1265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029869863778683454,
+      "loss": 0.0063,
+      "step": 1266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986960885505091,
+      "loss": 0.0337,
+      "step": 1267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029869353683067827,
+      "loss": 0.0316,
+      "step": 1268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029869098262738474,
+      "loss": 0.0259,
+      "step": 1269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986884259406711,
+      "loss": 0.0141,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029868586677058004,
+      "loss": 0.0101,
+      "step": 1271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986833051171544,
+      "loss": 0.0599,
+      "step": 1272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029868074098043695,
+      "loss": 0.0134,
+      "step": 1273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986781743604704,
+      "loss": 0.0273,
+      "step": 1274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029867560525729766,
+      "loss": 0.0293,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986730336709617,
+      "loss": 0.0296,
+      "step": 1276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029867045960150547,
+      "loss": 0.0092,
+      "step": 1277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986678830489719,
+      "loss": 0.0254,
+      "step": 1278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029866530401340407,
+      "loss": 0.026,
+      "step": 1279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000298662722494845,
+      "loss": 0.0138,
+      "step": 1280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029866013849333785,
+      "loss": 0.0288,
+      "step": 1281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986575520089258,
+      "loss": 0.02,
+      "step": 1282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000298654963041652,
+      "loss": 0.0361,
+      "step": 1283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986523715915597,
+      "loss": 0.0305,
+      "step": 1284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986497776586922,
+      "loss": 0.0215,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029864718124309283,
+      "loss": 0.0493,
+      "step": 1286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029864458234480496,
+      "loss": 0.0141,
+      "step": 1287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029864198096387197,
+      "loss": 0.0228,
+      "step": 1288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986393771003373,
+      "loss": 0.0671,
+      "step": 1289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029863677075424446,
+      "loss": 0.0298,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000298634161925637,
+      "loss": 0.0827,
+      "step": 1291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986315506145584,
+      "loss": 0.0421,
+      "step": 1292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986289368210524,
+      "loss": 0.0117,
+      "step": 1293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986263205451626,
+      "loss": 0.0109,
+      "step": 1294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029862370178693263,
+      "loss": 0.0291,
+      "step": 1295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986210805464064,
+      "loss": 0.0339,
+      "step": 1296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986184568236275,
+      "loss": 0.0577,
+      "step": 1297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029861583061863985,
+      "loss": 0.0117,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029861320193148736,
+      "loss": 0.0189,
+      "step": 1299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986105707622138,
+      "loss": 0.0333,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986079371108632,
+      "loss": 0.0292,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002986053009774795,
+      "loss": 0.0243,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002986026623621068,
+      "loss": 0.0346,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029860002126478913,
+      "loss": 0.0302,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985973776855706,
+      "loss": 0.0041,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985947316244954,
+      "loss": 0.0194,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029859208308160765,
+      "loss": 0.0103,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985894320569516,
+      "loss": 0.0192,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985867785505716,
+      "loss": 0.0387,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985841225625119,
+      "loss": 0.0095,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029858146409281693,
+      "loss": 0.0461,
+      "step": 1311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029857880314153096,
+      "loss": 0.0577,
+      "step": 1312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029857613970869854,
+      "loss": 0.014,
+      "step": 1313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029857347379436415,
+      "loss": 0.0372,
+      "step": 1314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985708053985723,
+      "loss": 0.0409,
+      "step": 1315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029856813452136755,
+      "loss": 0.1102,
+      "step": 1316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029856546116279446,
+      "loss": 0.0495,
+      "step": 1317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985627853228978,
+      "loss": 0.0519,
+      "step": 1318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029856010700172217,
+      "loss": 0.0292,
+      "step": 1319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985574261993123,
+      "loss": 0.0071,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298554742915713,
+      "loss": 0.0399,
+      "step": 1321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985520571509691,
+      "loss": 0.0072,
+      "step": 1322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985493689051254,
+      "loss": 0.0205,
+      "step": 1323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029854667817822686,
+      "loss": 0.0132,
+      "step": 1324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985439849703184,
+      "loss": 0.0131,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029854128928144503,
+      "loss": 0.0184,
+      "step": 1326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029853859111165164,
+      "loss": 0.0111,
+      "step": 1327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029853589046098343,
+      "loss": 0.0244,
+      "step": 1328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985331873294855,
+      "loss": 0.036,
+      "step": 1329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029853048171720294,
+      "loss": 0.0202,
+      "step": 1330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298527773624181,
+      "loss": 0.0321,
+      "step": 1331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985250630504648,
+      "loss": 0.0417,
+      "step": 1332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985223499960997,
+      "loss": 0.0431,
+      "step": 1333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029851963446113106,
+      "loss": 0.0254,
+      "step": 1334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029851691644560416,
+      "loss": 0.0138,
+      "step": 1335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985141959495644,
+      "loss": 0.0451,
+      "step": 1336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985114729730572,
+      "loss": 0.0059,
+      "step": 1337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985087475161281,
+      "loss": 0.0234,
+      "step": 1338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029850601957882254,
+      "loss": 0.0539,
+      "step": 1339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029850328916118614,
+      "loss": 0.0205,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985005562632645,
+      "loss": 0.0486,
+      "step": 1341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984978208851033,
+      "loss": 0.0182,
+      "step": 1342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984950830267482,
+      "loss": 0.034,
+      "step": 1343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984923426882448,
+      "loss": 0.0216,
+      "step": 1344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298489599869639,
+      "loss": 0.0305,
+      "step": 1345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984868545709767,
+      "loss": 0.0512,
+      "step": 1346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984841067923035,
+      "loss": 0.0515,
+      "step": 1347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029848135653366556,
+      "loss": 0.02,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029847860379510855,
+      "loss": 0.0335,
+      "step": 1349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984758485766787,
+      "loss": 0.0187,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029847309087842187,
+      "loss": 0.0654,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029847033070038413,
+      "loss": 0.0547,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029846756804261166,
+      "loss": 0.0224,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029846480290515054,
+      "loss": 0.0362,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029846203528804694,
+      "loss": 0.027,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984592651913471,
+      "loss": 0.0383,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029845649261509737,
+      "loss": 0.0344,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029845371755934393,
+      "loss": 0.0434,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029845094002413315,
+      "loss": 0.0225,
+      "step": 1359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984481600095115,
+      "loss": 0.0183,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029844537751552534,
+      "loss": 0.0086,
+      "step": 1361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984425925422211,
+      "loss": 0.0501,
+      "step": 1362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984398050896455,
+      "loss": 0.0139,
+      "step": 1363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029843701515784484,
+      "loss": 0.0335,
+      "step": 1364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029843422274686586,
+      "loss": 0.0266,
+      "step": 1365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029843142785675517,
+      "loss": 0.0279,
+      "step": 1366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984286304875595,
+      "loss": 0.0298,
+      "step": 1367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984258306393255,
+      "loss": 0.0103,
+      "step": 1368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984230283120999,
+      "loss": 0.0263,
+      "step": 1369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029842022350592967,
+      "loss": 0.033,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984174162208614,
+      "loss": 0.0636,
+      "step": 1371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984146064569422,
+      "loss": 0.0261,
+      "step": 1372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029841179421421897,
+      "loss": 0.0264,
+      "step": 1373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984089794927385,
+      "loss": 0.0081,
+      "step": 1374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984061622925481,
+      "loss": 0.01,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984033426136945,
+      "loss": 0.0199,
+      "step": 1376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000298400520456225,
+      "loss": 0.0177,
+      "step": 1377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983976958201867,
+      "loss": 0.0189,
+      "step": 1378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029839486870562675,
+      "loss": 0.0144,
+      "step": 1379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029839203911259237,
+      "loss": 0.0401,
+      "step": 1380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029838920704113085,
+      "loss": 0.0353,
+      "step": 1381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983863724912894,
+      "loss": 0.0134,
+      "step": 1382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029838353546311546,
+      "loss": 0.0407,
+      "step": 1383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029838069595665645,
+      "loss": 0.0494,
+      "step": 1384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029837785397195967,
+      "loss": 0.0435,
+      "step": 1385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029837500950907265,
+      "loss": 0.0172,
+      "step": 1386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029837216256804294,
+      "loss": 0.0114,
+      "step": 1387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000298369313148918,
+      "loss": 0.0516,
+      "step": 1388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029836646125174545,
+      "loss": 0.0359,
+      "step": 1389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000298363606876573,
+      "loss": 0.0474,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029836075002344814,
+      "loss": 0.0493,
+      "step": 1391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983578906924188,
+      "loss": 0.0045,
+      "step": 1392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983550288835326,
+      "loss": 0.0444,
+      "step": 1393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029835216459683745,
+      "loss": 0.027,
+      "step": 1394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029834929783238104,
+      "loss": 0.0436,
+      "step": 1395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029834642859021134,
+      "loss": 0.0295,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029834355687037627,
+      "loss": 0.0494,
+      "step": 1397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983406826729238,
+      "loss": 0.0213,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983378059979019,
+      "loss": 0.0171,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983349268453586,
+      "loss": 0.0494,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029833204521534205,
+      "loss": 0.0207,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029832916110790036,
+      "loss": 0.0287,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983262745230817,
+      "loss": 0.0514,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983233854609342,
+      "loss": 0.0141,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029832049392150626,
+      "loss": 0.027,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029831759990484605,
+      "loss": 0.0231,
+      "step": 1406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029831470341100196,
+      "loss": 0.0249,
+      "step": 1407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029831180444002235,
+      "loss": 0.0055,
+      "step": 1408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983089029919557,
+      "loss": 0.0196,
+      "step": 1409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029830599906685036,
+      "loss": 0.0265,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983030926647549,
+      "loss": 0.0097,
+      "step": 1411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029830018378571786,
+      "loss": 0.0242,
+      "step": 1412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982972724297878,
+      "loss": 0.0275,
+      "step": 1413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029829435859701336,
+      "loss": 0.022,
+      "step": 1414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982914422874432,
+      "loss": 0.0863,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982885235011261,
+      "loss": 0.0245,
+      "step": 1416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982856022381107,
+      "loss": 0.0409,
+      "step": 1417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982826784984458,
+      "loss": 0.0291,
+      "step": 1418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982797522821803,
+      "loss": 0.0108,
+      "step": 1419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029827682358936304,
+      "loss": 0.0227,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982738924200429,
+      "loss": 0.0482,
+      "step": 1421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029827095877426896,
+      "loss": 0.015,
+      "step": 1422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029826802265209,
+      "loss": 0.0363,
+      "step": 1423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029826508405355527,
+      "loss": 0.0261,
+      "step": 1424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029826214297871376,
+      "loss": 0.0495,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982591994276146,
+      "loss": 0.0327,
+      "step": 1426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982562534003069,
+      "loss": 0.0194,
+      "step": 1427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029825330489684,
+      "loss": 0.0235,
+      "step": 1428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029825035391726297,
+      "loss": 0.0288,
+      "step": 1429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982474004616252,
+      "loss": 0.0165,
+      "step": 1430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029824444452997606,
+      "loss": 0.0318,
+      "step": 1431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029824148612236483,
+      "loss": 0.0107,
+      "step": 1432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029823852523884096,
+      "loss": 0.018,
+      "step": 1433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982355618794539,
+      "loss": 0.0197,
+      "step": 1434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982325960442531,
+      "loss": 0.0271,
+      "step": 1435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029822962773328826,
+      "loss": 0.0198,
+      "step": 1436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982266569466087,
+      "loss": 0.0359,
+      "step": 1437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982236836842643,
+      "loss": 0.0508,
+      "step": 1438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029822070794630447,
+      "loss": 0.0353,
+      "step": 1439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029821772973277907,
+      "loss": 0.016,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029821474904373783,
+      "loss": 0.0372,
+      "step": 1441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029821176587923047,
+      "loss": 0.0227,
+      "step": 1442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029820878023930687,
+      "loss": 0.0114,
+      "step": 1443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029820579212401683,
+      "loss": 0.0141,
+      "step": 1444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982028015334104,
+      "loss": 0.0233,
+      "step": 1445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029819980846753735,
+      "loss": 0.0386,
+      "step": 1446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002981968129264478,
+      "loss": 0.009,
+      "step": 1447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002981938149101917,
+      "loss": 0.0345,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029819081441881916,
+      "loss": 0.0494,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029818781145238033,
+      "loss": 0.0185,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029818480601092527,
+      "loss": 0.0584,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981817980945043,
+      "loss": 0.0282,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981787877031675,
+      "loss": 0.0172,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981757748369653,
+      "loss": 0.0354,
+      "step": 1454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029817275949594793,
+      "loss": 0.046,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981697416801658,
+      "loss": 0.0547,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029816672138966924,
+      "loss": 0.0143,
+      "step": 1457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029816369862450883,
+      "loss": 0.0542,
+      "step": 1458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029816067338473486,
+      "loss": 0.033,
+      "step": 1459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981576456703981,
+      "loss": 0.0339,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029815461548154894,
+      "loss": 0.0382,
+      "step": 1461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029815158281823803,
+      "loss": 0.0323,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029814854768051604,
+      "loss": 0.0315,
+      "step": 1463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029814551006843366,
+      "loss": 0.0151,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981424699820416,
+      "loss": 0.0366,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029813942742139065,
+      "loss": 0.0071,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981363823865317,
+      "loss": 0.0094,
+      "step": 1467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029813333487751543,
+      "loss": 0.0067,
+      "step": 1468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981302848943929,
+      "loss": 0.0408,
+      "step": 1469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029812723243721496,
+      "loss": 0.0238,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981241775060327,
+      "loss": 0.0168,
+      "step": 1471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000298121120100897,
+      "loss": 0.0039,
+      "step": 1472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000298118060221859,
+      "loss": 0.0145,
+      "step": 1473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029811499786896984,
+      "loss": 0.0388,
+      "step": 1474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029811193304228064,
+      "loss": 0.0263,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981088657418425,
+      "loss": 0.0071,
+      "step": 1476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981057959677068,
+      "loss": 0.0349,
+      "step": 1477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981027237199247,
+      "loss": 0.044,
+      "step": 1478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980996489985476,
+      "loss": 0.009,
+      "step": 1479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980965718036267,
+      "loss": 0.0547,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029809349213521364,
+      "loss": 0.0317,
+      "step": 1481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029809040999335967,
+      "loss": 0.0191,
+      "step": 1482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029808732537811626,
+      "loss": 0.0403,
+      "step": 1483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029808423828953497,
+      "loss": 0.0272,
+      "step": 1484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029808114872766745,
+      "loss": 0.0221,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029807805669256515,
+      "loss": 0.0318,
+      "step": 1486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029807496218427985,
+      "loss": 0.0082,
+      "step": 1487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029807186520286314,
+      "loss": 0.0185,
+      "step": 1488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029806876574836675,
+      "loss": 0.0248,
+      "step": 1489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980656638208425,
+      "loss": 0.03,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029806255942034217,
+      "loss": 0.0439,
+      "step": 1491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980594525469176,
+      "loss": 0.014,
+      "step": 1492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980563432006207,
+      "loss": 0.0078,
+      "step": 1493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980532313815034,
+      "loss": 0.0527,
+      "step": 1494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029805011708961763,
+      "loss": 0.0084,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029804700032501553,
+      "loss": 0.0228,
+      "step": 1496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000298043881087749,
+      "loss": 0.0305,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029804075937787017,
+      "loss": 0.0264,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029803763519543127,
+      "loss": 0.0283,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980345085404844,
+      "loss": 0.0353,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029803137941308177,
+      "loss": 0.0166,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029802824781327575,
+      "loss": 0.0422,
+      "step": 1502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029802511374111846,
+      "loss": 0.034,
+      "step": 1503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029802197719666244,
+      "loss": 0.0542,
+      "step": 1504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029801883817996,
+      "loss": 0.0132,
+      "step": 1505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980156966910635,
+      "loss": 0.035,
+      "step": 1506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980125527300255,
+      "loss": 0.025,
+      "step": 1507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029800940629689843,
+      "loss": 0.0503,
+      "step": 1508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029800625739173495,
+      "loss": 0.0536,
+      "step": 1509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980031060145876,
+      "loss": 0.0111,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029799995216550895,
+      "loss": 0.0048,
+      "step": 1511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029799679584455174,
+      "loss": 0.0303,
+      "step": 1512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979936370517687,
+      "loss": 0.0299,
+      "step": 1513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979904757872125,
+      "loss": 0.0225,
+      "step": 1514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979873120509361,
+      "loss": 0.0215,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979841458429922,
+      "loss": 0.0042,
+      "step": 1516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979809771634337,
+      "loss": 0.0071,
+      "step": 1517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029797780601231353,
+      "loss": 0.0482,
+      "step": 1518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029797463238968473,
+      "loss": 0.0393,
+      "step": 1519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979714562956002,
+      "loss": 0.0562,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029796827773011307,
+      "loss": 0.0258,
+      "step": 1521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029796509669327635,
+      "loss": 0.0647,
+      "step": 1522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029796191318514324,
+      "loss": 0.0259,
+      "step": 1523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979587272057669,
+      "loss": 0.0136,
+      "step": 1524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979555387552005,
+      "loss": 0.0164,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029795234783349734,
+      "loss": 0.0154,
+      "step": 1526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979491544407107,
+      "loss": 0.0256,
+      "step": 1527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029794595857689384,
+      "loss": 0.0067,
+      "step": 1528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979427602421002,
+      "loss": 0.0022,
+      "step": 1529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979395594363833,
+      "loss": 0.0664,
+      "step": 1530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979363561597964,
+      "loss": 0.0213,
+      "step": 1531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979331504123932,
+      "loss": 0.016,
+      "step": 1532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979299421942271,
+      "loss": 0.0474,
+      "step": 1533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979267315053517,
+      "loss": 0.0245,
+      "step": 1534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979235183458207,
+      "loss": 0.1097,
+      "step": 1535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029792030271568766,
+      "loss": 0.0092,
+      "step": 1536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979170846150064,
+      "loss": 0.0283,
+      "step": 1537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979138640438306,
+      "loss": 0.0703,
+      "step": 1538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029791064100221406,
+      "loss": 0.022,
+      "step": 1539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979074154902106,
+      "loss": 0.0399,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979041875078741,
+      "loss": 0.0532,
+      "step": 1541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029790095705525853,
+      "loss": 0.0081,
+      "step": 1542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029789772413241775,
+      "loss": 0.0297,
+      "step": 1543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029789448873940576,
+      "loss": 0.0081,
+      "step": 1544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002978912508762767,
+      "loss": 0.0109,
+      "step": 1545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002978880105430846,
+      "loss": 0.0364,
+      "step": 1546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029788476773988354,
+      "loss": 0.0206,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978815224667277,
+      "loss": 0.0434,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978782747236713,
+      "loss": 0.0036,
+      "step": 1549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029787502451076854,
+      "loss": 0.0161,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978717718280738,
+      "loss": 0.0148,
+      "step": 1551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978685166756413,
+      "loss": 0.0128,
+      "step": 1552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029786525905352545,
+      "loss": 0.0154,
+      "step": 1553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029786199896178066,
+      "loss": 0.032,
+      "step": 1554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978587364004614,
+      "loss": 0.0249,
+      "step": 1555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029785547136962213,
+      "loss": 0.0218,
+      "step": 1556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029785220386931735,
+      "loss": 0.0212,
+      "step": 1557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978489338996017,
+      "loss": 0.0062,
+      "step": 1558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978456614605298,
+      "loss": 0.0255,
+      "step": 1559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029784238655215626,
+      "loss": 0.0158,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978391091745357,
+      "loss": 0.0512,
+      "step": 1561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978358293277231,
+      "loss": 0.0159,
+      "step": 1562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029783254701177305,
+      "loss": 0.0361,
+      "step": 1563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978292622267404,
+      "loss": 0.0428,
+      "step": 1564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029782597497267994,
+      "loss": 0.0231,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029782268524964673,
+      "loss": 0.0434,
+      "step": 1566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029781939305769566,
+      "loss": 0.0225,
+      "step": 1567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978160983968817,
+      "loss": 0.046,
+      "step": 1568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029781280126725985,
+      "loss": 0.017,
+      "step": 1569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029780950166888515,
+      "loss": 0.0375,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029780619960181285,
+      "loss": 0.0167,
+      "step": 1571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029780289506609796,
+      "loss": 0.0375,
+      "step": 1572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977995880617957,
+      "loss": 0.0371,
+      "step": 1573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029779627858896144,
+      "loss": 0.0209,
+      "step": 1574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029779296664765023,
+      "loss": 0.0183,
+      "step": 1575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029778965223791756,
+      "loss": 0.0445,
+      "step": 1576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977863353598187,
+      "loss": 0.0362,
+      "step": 1577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029778301601340906,
+      "loss": 0.0122,
+      "step": 1578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977796941987441,
+      "loss": 0.0203,
+      "step": 1579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977763699158793,
+      "loss": 0.0664,
+      "step": 1580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977730431648702,
+      "loss": 0.0183,
+      "step": 1581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977697139457723,
+      "loss": 0.0422,
+      "step": 1582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977663822586413,
+      "loss": 0.0331,
+      "step": 1583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029776304810353276,
+      "loss": 0.0257,
+      "step": 1584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977597114805023,
+      "loss": 0.0478,
+      "step": 1585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029775637238960585,
+      "loss": 0.0278,
+      "step": 1586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000297753030830899,
+      "loss": 0.0227,
+      "step": 1587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977496868044377,
+      "loss": 0.0164,
+      "step": 1588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977463403102777,
+      "loss": 0.009,
+      "step": 1589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029774299134847497,
+      "loss": 0.0272,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977396399190853,
+      "loss": 0.0185,
+      "step": 1591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977362860221649,
+      "loss": 0.0446,
+      "step": 1592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029773292965776954,
+      "loss": 0.0142,
+      "step": 1593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977295708259554,
+      "loss": 0.0255,
+      "step": 1594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029772620952677863,
+      "loss": 0.0595,
+      "step": 1595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977228457602953,
+      "loss": 0.0544,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977194795265615,
+      "loss": 0.0147,
+      "step": 1597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029771611082563367,
+      "loss": 0.0281,
+      "step": 1598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977127396575679,
+      "loss": 0.0759,
+      "step": 1599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029770936602242055,
+      "loss": 0.0243,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000297705989920248,
+      "loss": 0.0217,
+      "step": 1601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029770261135110657,
+      "loss": 0.0103,
+      "step": 1602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029769923031505274,
+      "loss": 0.0097,
+      "step": 1603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976958468121429,
+      "loss": 0.0354,
+      "step": 1604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976924608424337,
+      "loss": 0.0256,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976890724059816,
+      "loss": 0.0298,
+      "step": 1606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976856815028432,
+      "loss": 0.0184,
+      "step": 1607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976822881330751,
+      "loss": 0.0078,
+      "step": 1608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029767889229673406,
+      "loss": 0.0145,
+      "step": 1609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976754939938768,
+      "loss": 0.0253,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029767209322456,
+      "loss": 0.0119,
+      "step": 1611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029766868998884043,
+      "loss": 0.0176,
+      "step": 1612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976652842867751,
+      "loss": 0.0859,
+      "step": 1613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029766187611842066,
+      "loss": 0.0173,
+      "step": 1614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976584654838342,
+      "loss": 0.0417,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029765505238307273,
+      "loss": 0.0207,
+      "step": 1616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976516368161931,
+      "loss": 0.0434,
+      "step": 1617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976482187832524,
+      "loss": 0.0124,
+      "step": 1618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976447982843078,
+      "loss": 0.0309,
+      "step": 1619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976413753194163,
+      "loss": 0.0314,
+      "step": 1620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029763794988863515,
+      "loss": 0.0154,
+      "step": 1621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976345219920216,
+      "loss": 0.0155,
+      "step": 1622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029763109162963283,
+      "loss": 0.0272,
+      "step": 1623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029762765880152615,
+      "loss": 0.0274,
+      "step": 1624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029762422350775887,
+      "loss": 0.0113,
+      "step": 1625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029762078574838846,
+      "loss": 0.0329,
+      "step": 1626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029761734552347227,
+      "loss": 0.0206,
+      "step": 1627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976139028330677,
+      "loss": 0.0061,
+      "step": 1628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029761045767723234,
+      "loss": 0.0265,
+      "step": 1629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029760701005602375,
+      "loss": 0.0336,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029760355996949944,
+      "loss": 0.025,
+      "step": 1631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029760010741771705,
+      "loss": 0.0149,
+      "step": 1632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975966524007342,
+      "loss": 0.042,
+      "step": 1633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975931949186087,
+      "loss": 0.008,
+      "step": 1634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029758973497139826,
+      "loss": 0.0166,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975862725591606,
+      "loss": 0.0373,
+      "step": 1636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975828076819536,
+      "loss": 0.0078,
+      "step": 1637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029757934033983516,
+      "loss": 0.019,
+      "step": 1638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029757587053286314,
+      "loss": 0.0538,
+      "step": 1639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029757239826109556,
+      "loss": 0.0108,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975689235245903,
+      "loss": 0.028,
+      "step": 1641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975654463234055,
+      "loss": 0.0065,
+      "step": 1642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975619666575992,
+      "loss": 0.0295,
+      "step": 1643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975584845272295,
+      "loss": 0.0229,
+      "step": 1644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975549999323546,
+      "loss": 0.0127,
+      "step": 1645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029755151287303263,
+      "loss": 0.0396,
+      "step": 1646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975480233493219,
+      "loss": 0.0316,
+      "step": 1647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029754453136128067,
+      "loss": 0.016,
+      "step": 1648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975410369089672,
+      "loss": 0.0242,
+      "step": 1649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029753753999243993,
+      "loss": 0.0177,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975340406117573,
+      "loss": 0.0312,
+      "step": 1651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029753053876697765,
+      "loss": 0.0136,
+      "step": 1652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029752703445815954,
+      "loss": 0.0235,
+      "step": 1653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029752352768536145,
+      "loss": 0.0296,
+      "step": 1654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000297520018448642,
+      "loss": 0.0263,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975165067480598,
+      "loss": 0.0248,
+      "step": 1656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029751299258367345,
+      "loss": 0.0106,
+      "step": 1657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975094759555417,
+      "loss": 0.0533,
+      "step": 1658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029750595686372323,
+      "loss": 0.0178,
+      "step": 1659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029750243530827684,
+      "loss": 0.0355,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974989112892614,
+      "loss": 0.0144,
+      "step": 1661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974953848067356,
+      "loss": 0.0049,
+      "step": 1662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974918558607586,
+      "loss": 0.0223,
+      "step": 1663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029748832445138914,
+      "loss": 0.0097,
+      "step": 1664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974847905786862,
+      "loss": 0.0277,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029748125424270896,
+      "loss": 0.0231,
+      "step": 1666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974777154435163,
+      "loss": 0.0177,
+      "step": 1667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974741741811674,
+      "loss": 0.0542,
+      "step": 1668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029747063045572147,
+      "loss": 0.0449,
+      "step": 1669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029746708426723765,
+      "loss": 0.0146,
+      "step": 1670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029746353561577513,
+      "loss": 0.0321,
+      "step": 1671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029745998450139325,
+      "loss": 0.0177,
+      "step": 1672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974564309241512,
+      "loss": 0.0313,
+      "step": 1673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029745287488410843,
+      "loss": 0.0424,
+      "step": 1674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974493163813244,
+      "loss": 0.0131,
+      "step": 1675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029744575541585836,
+      "loss": 0.0187,
+      "step": 1676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974421919877699,
+      "loss": 0.0378,
+      "step": 1677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974386260971185,
+      "loss": 0.0134,
+      "step": 1678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974350577439638,
+      "loss": 0.0082,
+      "step": 1679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029743148692836527,
+      "loss": 0.027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029742791365038263,
+      "loss": 0.0092,
+      "step": 1681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974243379100756,
+      "loss": 0.0192,
+      "step": 1682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974207597075038,
+      "loss": 0.0253,
+      "step": 1683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000297417179042727,
+      "loss": 0.0092,
+      "step": 1684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029741359591580504,
+      "loss": 0.0101,
+      "step": 1685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974100103267978,
+      "loss": 0.0228,
+      "step": 1686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029740642227576514,
+      "loss": 0.0293,
+      "step": 1687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029740283176276695,
+      "loss": 0.0498,
+      "step": 1688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002973992387878633,
+      "loss": 0.0321,
+      "step": 1689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000297395643351114,
+      "loss": 0.0504,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002973920454525793,
+      "loss": 0.016,
+      "step": 1691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002973884450923193,
+      "loss": 0.0241,
+      "step": 1692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029738484227039393,
+      "loss": 0.0182,
+      "step": 1693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029738123698686345,
+      "loss": 0.062,
+      "step": 1694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973776292417882,
+      "loss": 0.0212,
+      "step": 1695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029737401903522835,
+      "loss": 0.0029,
+      "step": 1696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973704063672442,
+      "loss": 0.0936,
+      "step": 1697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029736679123789603,
+      "loss": 0.0266,
+      "step": 1698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973631736472443,
+      "loss": 0.0236,
+      "step": 1699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973595535953494,
+      "loss": 0.0142,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973559310822718,
+      "loss": 0.0302,
+      "step": 1701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029735230610807196,
+      "loss": 0.0215,
+      "step": 1702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973486786728105,
+      "loss": 0.0473,
+      "step": 1703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000297345048776548,
+      "loss": 0.0091,
+      "step": 1704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029734141641934493,
+      "loss": 0.0395,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973377816012622,
+      "loss": 0.0325,
+      "step": 1706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029733414432236033,
+      "loss": 0.0212,
+      "step": 1707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029733050458270017,
+      "loss": 0.0346,
+      "step": 1708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973268623823425,
+      "loss": 0.0057,
+      "step": 1709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029732321772134806,
+      "loss": 0.0229,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973195705997778,
+      "loss": 0.0296,
+      "step": 1711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973159210176927,
+      "loss": 0.0296,
+      "step": 1712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973122689751536,
+      "loss": 0.0562,
+      "step": 1713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973086144722215,
+      "loss": 0.0406,
+      "step": 1714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029730495750895756,
+      "loss": 0.0594,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973012980854227,
+      "loss": 0.0226,
+      "step": 1716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972976362016781,
+      "loss": 0.0229,
+      "step": 1717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029729397185778496,
+      "loss": 0.0223,
+      "step": 1718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972903050538045,
+      "loss": 0.0261,
+      "step": 1719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029728663578979783,
+      "loss": 0.0555,
+      "step": 1720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972829640658264,
+      "loss": 0.0392,
+      "step": 1721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029727928988195135,
+      "loss": 0.017,
+      "step": 1722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029727561323823423,
+      "loss": 0.0094,
+      "step": 1723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029727193413473635,
+      "loss": 0.0143,
+      "step": 1724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029726825257151915,
+      "loss": 0.0199,
+      "step": 1725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972645685486442,
+      "loss": 0.021,
+      "step": 1726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972608820661729,
+      "loss": 0.0303,
+      "step": 1727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000297257193124167,
+      "loss": 0.0175,
+      "step": 1728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972535017226879,
+      "loss": 0.0511,
+      "step": 1729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972498078617974,
+      "loss": 0.0167,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029724611154155717,
+      "loss": 0.0575,
+      "step": 1731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972424127620289,
+      "loss": 0.0379,
+      "step": 1732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029723871152327445,
+      "loss": 0.0417,
+      "step": 1733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029723500782535557,
+      "loss": 0.0204,
+      "step": 1734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029723130166833404,
+      "loss": 0.0158,
+      "step": 1735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029722759305227196,
+      "loss": 0.0368,
+      "step": 1736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029722388197723115,
+      "loss": 0.0165,
+      "step": 1737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029722016844327355,
+      "loss": 0.0167,
+      "step": 1738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029721645245046126,
+      "loss": 0.0042,
+      "step": 1739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029721273399885636,
+      "loss": 0.0682,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029720901308852093,
+      "loss": 0.0282,
+      "step": 1741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000297205289719517,
+      "loss": 0.0268,
+      "step": 1742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000297201563891907,
+      "loss": 0.0285,
+      "step": 1743
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971978356057529,
+      "loss": 0.0153,
+      "step": 1744
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029719410486111716,
+      "loss": 0.0325,
+      "step": 1745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297190371658062,
+      "loss": 0.0164,
+      "step": 1746
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971866359966498,
+      "loss": 0.0348,
+      "step": 1747
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029718289787694294,
+      "loss": 0.0217,
+      "step": 1748
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971791572990039,
+      "loss": 0.0079,
+      "step": 1749
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029717541426289504,
+      "loss": 0.0734,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297171668768679,
+      "loss": 0.0463,
+      "step": 1751
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029716792081641827,
+      "loss": 0.0374,
+      "step": 1752
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029716417040617543,
+      "loss": 0.0328,
+      "step": 1753
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971604175380132,
+      "loss": 0.0193,
+      "step": 1754
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029715666221199414,
+      "loss": 0.034,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029715290442818113,
+      "loss": 0.0358,
+      "step": 1756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029714914418663683,
+      "loss": 0.0299,
+      "step": 1757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297145381487424,
+      "loss": 0.0801,
+      "step": 1758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971416163306056,
+      "loss": 0.0126,
+      "step": 1759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029713784871624447,
+      "loss": 0.0489,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971340786444035,
+      "loss": 0.0133,
+      "step": 1761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029713030611514565,
+      "loss": 0.0115,
+      "step": 1762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029712653112853404,
+      "loss": 0.0352,
+      "step": 1763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029712275368463156,
+      "loss": 0.0381,
+      "step": 1764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971189737835014,
+      "loss": 0.0134,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029711519142520666,
+      "loss": 0.0305,
+      "step": 1766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971114066098105,
+      "loss": 0.0106,
+      "step": 1767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029710761933737627,
+      "loss": 0.0227,
+      "step": 1768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297103829607967,
+      "loss": 0.0055,
+      "step": 1769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029710003742164614,
+      "loss": 0.0431,
+      "step": 1770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297096242778477,
+      "loss": 0.0171,
+      "step": 1771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970924456785229,
+      "loss": 0.0086,
+      "step": 1772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970886461218473,
+      "loss": 0.0214,
+      "step": 1773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970848441085137,
+      "loss": 0.0306,
+      "step": 1774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029708103963858553,
+      "loss": 0.0301,
+      "step": 1775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970772327121264,
+      "loss": 0.0502,
+      "step": 1776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970734233291998,
+      "loss": 0.0674,
+      "step": 1777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029706961148986944,
+      "loss": 0.0315,
+      "step": 1778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297065797194199,
+      "loss": 0.0087,
+      "step": 1779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029706198044225206,
+      "loss": 0.0291,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970581612340925,
+      "loss": 0.0386,
+      "step": 1781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297054339569784,
+      "loss": 0.0156,
+      "step": 1782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970505154493905,
+      "loss": 0.0065,
+      "step": 1783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970466888729758,
+      "loss": 0.0555,
+      "step": 1784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029704285984060385,
+      "loss": 0.0242,
+      "step": 1785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029703902835233857,
+      "loss": 0.0789,
+      "step": 1786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970351944082439,
+      "loss": 0.0428,
+      "step": 1787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029703135800838396,
+      "loss": 0.0176,
+      "step": 1788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029702751915282284,
+      "loss": 0.0196,
+      "step": 1789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970236778416246,
+      "loss": 0.0548,
+      "step": 1790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029701983407485344,
+      "loss": 0.0098,
+      "step": 1791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970159878525735,
+      "loss": 0.0104,
+      "step": 1792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029701213917484904,
+      "loss": 0.0844,
+      "step": 1793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029700828804174435,
+      "loss": 0.0219,
+      "step": 1794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029700443445332376,
+      "loss": 0.0183,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002970005784096516,
+      "loss": 0.025,
+      "step": 1796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029699671991079233,
+      "loss": 0.0332,
+      "step": 1797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029699285895681035,
+      "loss": 0.0542,
+      "step": 1798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969889955477702,
+      "loss": 0.0171,
+      "step": 1799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969851296837363,
+      "loss": 0.0165,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969812613647733,
+      "loss": 0.0303,
+      "step": 1801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969773905909458,
+      "loss": 0.0486,
+      "step": 1802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969735173623184,
+      "loss": 0.0259,
+      "step": 1803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029696964167895586,
+      "loss": 0.0089,
+      "step": 1804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029696576354092287,
+      "loss": 0.0165,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029696188294828425,
+      "loss": 0.0693,
+      "step": 1806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969579999011047,
+      "loss": 0.0104,
+      "step": 1807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969541143994492,
+      "loss": 0.0208,
+      "step": 1808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029695022644338263,
+      "loss": 0.0444,
+      "step": 1809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969463360329698,
+      "loss": 0.0347,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029694244316827586,
+      "loss": 0.0542,
+      "step": 1811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969385478493657,
+      "loss": 0.0564,
+      "step": 1812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969346500763044,
+      "loss": 0.0173,
+      "step": 1813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029693074984915715,
+      "loss": 0.0277,
+      "step": 1814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029692684716798895,
+      "loss": 0.0199,
+      "step": 1815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029692294203286515,
+      "loss": 0.0343,
+      "step": 1816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969190344438508,
+      "loss": 0.0152,
+      "step": 1817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029691512440101133,
+      "loss": 0.0338,
+      "step": 1818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969112119044119,
+      "loss": 0.0165,
+      "step": 1819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029690729695411793,
+      "loss": 0.0091,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029690337955019477,
+      "loss": 0.0415,
+      "step": 1821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968994596927079,
+      "loss": 0.0141,
+      "step": 1822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029689553738172276,
+      "loss": 0.031,
+      "step": 1823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968916126173048,
+      "loss": 0.0183,
+      "step": 1824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968876853995197,
+      "loss": 0.0446,
+      "step": 1825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029688375572843297,
+      "loss": 0.0205,
+      "step": 1826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968798236041103,
+      "loss": 0.0058,
+      "step": 1827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029687588902661724,
+      "loss": 0.0215,
+      "step": 1828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968719519960196,
+      "loss": 0.0328,
+      "step": 1829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029686801251238316,
+      "loss": 0.0162,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029686407057577367,
+      "loss": 0.0415,
+      "step": 1831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029686012618625695,
+      "loss": 0.0252,
+      "step": 1832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968561793438989,
+      "loss": 0.0134,
+      "step": 1833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029685223004876553,
+      "loss": 0.0559,
+      "step": 1834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968482783009226,
+      "loss": 0.0161,
+      "step": 1835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968443241004363,
+      "loss": 0.0023,
+      "step": 1836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968403674473726,
+      "loss": 0.0238,
+      "step": 1837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968364083417975,
+      "loss": 0.0127,
+      "step": 1838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968324467837773,
+      "loss": 0.0318,
+      "step": 1839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029682848277337804,
+      "loss": 0.0176,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000296824516310666,
+      "loss": 0.031,
+      "step": 1841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029682054739570726,
+      "loss": 0.0158,
+      "step": 1842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029681657602856836,
+      "loss": 0.0169,
+      "step": 1843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029681260220931543,
+      "loss": 0.0148,
+      "step": 1844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000296808625938015,
+      "loss": 0.031,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002968046472147333,
+      "loss": 0.0499,
+      "step": 1846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029680066603953695,
+      "loss": 0.0712,
+      "step": 1847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029679668241249234,
+      "loss": 0.0242,
+      "step": 1848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029679269633366606,
+      "loss": 0.0081,
+      "step": 1849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029678870780312463,
+      "loss": 0.0205,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967847168209347,
+      "loss": 0.0427,
+      "step": 1851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967807233871629,
+      "loss": 0.0441,
+      "step": 1852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029677672750187597,
+      "loss": 0.0123,
+      "step": 1853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967727291651407,
+      "loss": 0.0159,
+      "step": 1854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967687283770237,
+      "loss": 0.0183,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029676472513759193,
+      "loss": 0.0232,
+      "step": 1856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967607194469122,
+      "loss": 0.0409,
+      "step": 1857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029675671130505144,
+      "loss": 0.0072,
+      "step": 1858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967527007120766,
+      "loss": 0.0434,
+      "step": 1859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967486876680546,
+      "loss": 0.015,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029674467217305256,
+      "loss": 0.0269,
+      "step": 1861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029674065422713745,
+      "loss": 0.0189,
+      "step": 1862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029673663383037645,
+      "loss": 0.0433,
+      "step": 1863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029673261098283666,
+      "loss": 0.0838,
+      "step": 1864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029672858568458526,
+      "loss": 0.0249,
+      "step": 1865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029672455793568957,
+      "loss": 0.0131,
+      "step": 1866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029672052773621675,
+      "loss": 0.0289,
+      "step": 1867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967164950862342,
+      "loss": 0.0109,
+      "step": 1868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967124599858092,
+      "loss": 0.0341,
+      "step": 1869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029670842243500925,
+      "loss": 0.0102,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029670438243390167,
+      "loss": 0.0174,
+      "step": 1871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029670033998255394,
+      "loss": 0.0064,
+      "step": 1872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029669629508103366,
+      "loss": 0.0107,
+      "step": 1873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966922477294083,
+      "loss": 0.0081,
+      "step": 1874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966881979277455,
+      "loss": 0.0338,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029668414567611296,
+      "loss": 0.0139,
+      "step": 1876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029668009097457825,
+      "loss": 0.0433,
+      "step": 1877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966760338232092,
+      "loss": 0.0552,
+      "step": 1878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966719742220735,
+      "loss": 0.0139,
+      "step": 1879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029666791217123896,
+      "loss": 0.0037,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966638476707734,
+      "loss": 0.0332,
+      "step": 1881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029665978072074473,
+      "loss": 0.0366,
+      "step": 1882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966557113212209,
+      "loss": 0.0171,
+      "step": 1883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966516394722699,
+      "loss": 0.0158,
+      "step": 1884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029664756517395964,
+      "loss": 0.0162,
+      "step": 1885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029664348842635825,
+      "loss": 0.0222,
+      "step": 1886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966394092295338,
+      "loss": 0.0297,
+      "step": 1887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966353275835544,
+      "loss": 0.0084,
+      "step": 1888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029663124348848827,
+      "loss": 0.0247,
+      "step": 1889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029662715694440354,
+      "loss": 0.0115,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029662306795136847,
+      "loss": 0.009,
+      "step": 1891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002966189765094515,
+      "loss": 0.0132,
+      "step": 1892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029661488261872084,
+      "loss": 0.0361,
+      "step": 1893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002966107862792449,
+      "loss": 0.0623,
+      "step": 1894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296606687491092,
+      "loss": 0.0133,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002966025862543307,
+      "loss": 0.0168,
+      "step": 1896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029659848256902946,
+      "loss": 0.0376,
+      "step": 1897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965943764352569,
+      "loss": 0.0198,
+      "step": 1898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965902678530815,
+      "loss": 0.0281,
+      "step": 1899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029658615682257196,
+      "loss": 0.0277,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029658204334379685,
+      "loss": 0.0501,
+      "step": 1901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029657792741682494,
+      "loss": 0.0131,
+      "step": 1902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296573809041725,
+      "loss": 0.0338,
+      "step": 1903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965696882185657,
+      "loss": 0.012,
+      "step": 1904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296565564947416,
+      "loss": 0.0051,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029656143922834465,
+      "loss": 0.0278,
+      "step": 1906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965573110614206,
+      "loss": 0.0793,
+      "step": 1907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965531804467129,
+      "loss": 0.0251,
+      "step": 1908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029654904738429035,
+      "loss": 0.0085,
+      "step": 1909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965449118742221,
+      "loss": 0.0454,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029654077391657723,
+      "loss": 0.025,
+      "step": 1911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029653663351142476,
+      "loss": 0.0273,
+      "step": 1912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296532490658834,
+      "loss": 0.02,
+      "step": 1913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296528345358874,
+      "loss": 0.0199,
+      "step": 1914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296524197611614,
+      "loss": 0.0193,
+      "step": 1915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965200474171234,
+      "loss": 0.0201,
+      "step": 1916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029651589477547134,
+      "loss": 0.0131,
+      "step": 1917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965117396867273,
+      "loss": 0.0068,
+      "step": 1918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965075821509607,
+      "loss": 0.0495,
+      "step": 1919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965034221682409,
+      "loss": 0.02,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029649925973863737,
+      "loss": 0.0129,
+      "step": 1921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029649509486221967,
+      "loss": 0.0229,
+      "step": 1922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029649092753905734,
+      "loss": 0.0112,
+      "step": 1923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029648675776922003,
+      "loss": 0.0117,
+      "step": 1924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029648258555277737,
+      "loss": 0.0551,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296478410889799,
+      "loss": 0.0199,
+      "step": 1926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964742337803547,
+      "loss": 0.0439,
+      "step": 1927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029647005422451416,
+      "loss": 0.0073,
+      "step": 1928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964658722223473,
+      "loss": 0.0366,
+      "step": 1929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029646168777392384,
+      "loss": 0.0168,
+      "step": 1930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964575008793137,
+      "loss": 0.0852,
+      "step": 1931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029645331153858694,
+      "loss": 0.009,
+      "step": 1932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029644911975181334,
+      "loss": 0.0159,
+      "step": 1933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296444925519063,
+      "loss": 0.0167,
+      "step": 1934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029644072884040597,
+      "loss": 0.0327,
+      "step": 1935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029643652971591243,
+      "loss": 0.0219,
+      "step": 1936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029643232814565237,
+      "loss": 0.0324,
+      "step": 1937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296428124129696,
+      "loss": 0.0307,
+      "step": 1938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029642391766811357,
+      "loss": 0.0381,
+      "step": 1939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002964197087609753,
+      "loss": 0.0098,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002964154974083515,
+      "loss": 0.0089,
+      "step": 1941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029641128361031254,
+      "loss": 0.0117,
+      "step": 1942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029640706736692876,
+      "loss": 0.0327,
+      "step": 1943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029640284867827064,
+      "loss": 0.0555,
+      "step": 1944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029639862754440855,
+      "loss": 0.0215,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029639440396541305,
+      "loss": 0.0127,
+      "step": 1946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029639017794135463,
+      "loss": 0.0205,
+      "step": 1947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000296385949472304,
+      "loss": 0.0373,
+      "step": 1948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029638171855833156,
+      "loss": 0.0081,
+      "step": 1949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963774851995082,
+      "loss": 0.0232,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029637324939590443,
+      "loss": 0.0188,
+      "step": 1951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963690111475912,
+      "loss": 0.0223,
+      "step": 1952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029636477045463913,
+      "loss": 0.0121,
+      "step": 1953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029636052731711913,
+      "loss": 0.0353,
+      "step": 1954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029635628173510205,
+      "loss": 0.0196,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029635203370865885,
+      "loss": 0.0442,
+      "step": 1956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029634778323786034,
+      "loss": 0.0172,
+      "step": 1957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963435303227777,
+      "loss": 0.0082,
+      "step": 1958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029633927496348174,
+      "loss": 0.0191,
+      "step": 1959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029633501716004373,
+      "loss": 0.0305,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963307569125347,
+      "loss": 0.0144,
+      "step": 1961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963264942210258,
+      "loss": 0.0642,
+      "step": 1962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029632222908558824,
+      "loss": 0.0115,
+      "step": 1963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029631796150629325,
+      "loss": 0.0183,
+      "step": 1964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963136914832121,
+      "loss": 0.0129,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963094190164161,
+      "loss": 0.0319,
+      "step": 1966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963051441059766,
+      "loss": 0.0322,
+      "step": 1967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963008667519651,
+      "loss": 0.0314,
+      "step": 1968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029629658695445296,
+      "loss": 0.0132,
+      "step": 1969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962923047135116,
+      "loss": 0.0539,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962880200292127,
+      "loss": 0.0128,
+      "step": 1971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962837329016277,
+      "loss": 0.0524,
+      "step": 1972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029627944333082814,
+      "loss": 0.0186,
+      "step": 1973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962751513168858,
+      "loss": 0.0187,
+      "step": 1974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962708568598723,
+      "loss": 0.0233,
+      "step": 1975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029626655995985946,
+      "loss": 0.0263,
+      "step": 1976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029626226061691893,
+      "loss": 0.0232,
+      "step": 1977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029625795883112254,
+      "loss": 0.0198,
+      "step": 1978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029625365460254216,
+      "loss": 0.0393,
+      "step": 1979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962493479312497,
+      "loss": 0.025,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029624503881731703,
+      "loss": 0.0562,
+      "step": 1981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029624072726081615,
+      "loss": 0.0283,
+      "step": 1982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029623641326181915,
+      "loss": 0.0262,
+      "step": 1983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029623209682039795,
+      "loss": 0.0058,
+      "step": 1984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962277779366247,
+      "loss": 0.0461,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029622345661057154,
+      "loss": 0.0261,
+      "step": 1986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029621913284231063,
+      "loss": 0.0364,
+      "step": 1987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962148066319142,
+      "loss": 0.0408,
+      "step": 1988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962104779794545,
+      "loss": 0.0363,
+      "step": 1989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029620614688500386,
+      "loss": 0.032,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002962018133486345,
+      "loss": 0.0064,
+      "step": 1991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029619747737041896,
+      "loss": 0.0481,
+      "step": 1992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961931389504296,
+      "loss": 0.0077,
+      "step": 1993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961887980887388,
+      "loss": 0.1338,
+      "step": 1994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029618445478541913,
+      "loss": 0.0204,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029618010904054314,
+      "loss": 0.0156,
+      "step": 1996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029617576085418334,
+      "loss": 0.0374,
+      "step": 1997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029617141022641243,
+      "loss": 0.0081,
+      "step": 1998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961670571573031,
+      "loss": 0.0108,
+      "step": 1999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000296162701646928,
+      "loss": 0.0218,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961583436953598,
+      "loss": 0.056,
+      "step": 2001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029615398330267145,
+      "loss": 0.0077,
+      "step": 2002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029614962046893567,
+      "loss": 0.0385,
+      "step": 2003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029614525519422533,
+      "loss": 0.0229,
+      "step": 2004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961408874786134,
+      "loss": 0.013,
+      "step": 2005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961365173221728,
+      "loss": 0.0191,
+      "step": 2006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029613214472497647,
+      "loss": 0.009,
+      "step": 2007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029612776968709746,
+      "loss": 0.0353,
+      "step": 2008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961233922086089,
+      "loss": 0.0309,
+      "step": 2009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961190122895838,
+      "loss": 0.0646,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029611462993009544,
+      "loss": 0.0102,
+      "step": 2011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029611024513021694,
+      "loss": 0.0384,
+      "step": 2012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961058578900215,
+      "loss": 0.0076,
+      "step": 2013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961014682095825,
+      "loss": 0.0426,
+      "step": 2014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029609707608897317,
+      "loss": 0.0327,
+      "step": 2015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960926815282669,
+      "loss": 0.0143,
+      "step": 2016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029608828452753707,
+      "loss": 0.026,
+      "step": 2017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029608388508685715,
+      "loss": 0.0372,
+      "step": 2018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960794832063006,
+      "loss": 0.031,
+      "step": 2019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960750788859409,
+      "loss": 0.0053,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029607067212585167,
+      "loss": 0.0448,
+      "step": 2021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029606626292610645,
+      "loss": 0.0486,
+      "step": 2022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029606185128677894,
+      "loss": 0.0343,
+      "step": 2023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029605743720794277,
+      "loss": 0.011,
+      "step": 2024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960530206896717,
+      "loss": 0.0199,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960486017320396,
+      "loss": 0.0198,
+      "step": 2026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960441803351201,
+      "loss": 0.0834,
+      "step": 2027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029603975649898705,
+      "loss": 0.0273,
+      "step": 2028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029603533022371447,
+      "loss": 0.0396,
+      "step": 2029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960309015093762,
+      "loss": 0.0111,
+      "step": 2030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960264703560462,
+      "loss": 0.0183,
+      "step": 2031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960220367637985,
+      "loss": 0.0265,
+      "step": 2032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029601760073270716,
+      "loss": 0.0129,
+      "step": 2033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029601316226284625,
+      "loss": 0.0213,
+      "step": 2034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960087213542899,
+      "loss": 0.0334,
+      "step": 2035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029600427800711233,
+      "loss": 0.0094,
+      "step": 2036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002959998322213877,
+      "loss": 0.0358,
+      "step": 2037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002959953839971903,
+      "loss": 0.0224,
+      "step": 2038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029599093333459433,
+      "loss": 0.0278,
+      "step": 2039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029598648023367426,
+      "loss": 0.0242,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959820246945044,
+      "loss": 0.0061,
+      "step": 2041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029597756671715915,
+      "loss": 0.0342,
+      "step": 2042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295973106301713,
+      "loss": 0.0067,
+      "step": 2043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959686434482404,
+      "loss": 0.0182,
+      "step": 2044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959641781568159,
+      "loss": 0.0059,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029595971042751417,
+      "loss": 0.008,
+      "step": 2046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959552402604097,
+      "loss": 0.0673,
+      "step": 2047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029595076765557725,
+      "loss": 0.0164,
+      "step": 2048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959462926130915,
+      "loss": 0.0137,
+      "step": 2049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959418151330271,
+      "loss": 0.0294,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959373352154589,
+      "loss": 0.0353,
+      "step": 2051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959328528604618,
+      "loss": 0.0533,
+      "step": 2052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959283680681105,
+      "loss": 0.0131,
+      "step": 2053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029592388083848003,
+      "loss": 0.0251,
+      "step": 2054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029591939117164527,
+      "loss": 0.0204,
+      "step": 2055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029591489906768135,
+      "loss": 0.012,
+      "step": 2056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029591040452666307,
+      "loss": 0.0285,
+      "step": 2057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959059075486656,
+      "loss": 0.054,
+      "step": 2058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029590140813376414,
+      "loss": 0.012,
+      "step": 2059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958969062820337,
+      "loss": 0.0338,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958924019935496,
+      "loss": 0.0247,
+      "step": 2061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958878952683869,
+      "loss": 0.0181,
+      "step": 2062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295883386106621,
+      "loss": 0.045,
+      "step": 2063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029587887450832715,
+      "loss": 0.0402,
+      "step": 2064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958743604735808,
+      "loss": 0.0264,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958698440024572,
+      "loss": 0.0587,
+      "step": 2066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029586532509503187,
+      "loss": 0.0285,
+      "step": 2067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029586080375138034,
+      "loss": 0.0046,
+      "step": 2068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295856279971578,
+      "loss": 0.0537,
+      "step": 2069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029585175375570046,
+      "loss": 0.0641,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029584722510382335,
+      "loss": 0.0205,
+      "step": 2071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958426940160222,
+      "loss": 0.0229,
+      "step": 2072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958381604923728,
+      "loss": 0.0373,
+      "step": 2073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029583362453295087,
+      "loss": 0.0315,
+      "step": 2074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958290861378321,
+      "loss": 0.041,
+      "step": 2075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958245453070923,
+      "loss": 0.0212,
+      "step": 2076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958200020408074,
+      "loss": 0.0107,
+      "step": 2077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029581545633905313,
+      "loss": 0.0169,
+      "step": 2078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029581090820190553,
+      "loss": 0.0215,
+      "step": 2079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958063576294405,
+      "loss": 0.0149,
+      "step": 2080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958018046217341,
+      "loss": 0.0061,
+      "step": 2081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029579724917886237,
+      "loss": 0.0156,
+      "step": 2082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957926913009013,
+      "loss": 0.0307,
+      "step": 2083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957881309879271,
+      "loss": 0.0262,
+      "step": 2084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029578356824001595,
+      "loss": 0.0158,
+      "step": 2085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295779003057244,
+      "loss": 0.0169,
+      "step": 2086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957744354396875,
+      "loss": 0.0177,
+      "step": 2087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029576986538742283,
+      "loss": 0.028,
+      "step": 2088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957652929005262,
+      "loss": 0.0109,
+      "step": 2089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295760717979074,
+      "loss": 0.0497,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957561406231427,
+      "loss": 0.0268,
+      "step": 2091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957515608328087,
+      "loss": 0.0221,
+      "step": 2092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957469786081485,
+      "loss": 0.0476,
+      "step": 2093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029574239394923867,
+      "loss": 0.0189,
+      "step": 2094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957378068561558,
+      "loss": 0.0084,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029573321732897633,
+      "loss": 0.0103,
+      "step": 2096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029572862536777713,
+      "loss": 0.034,
+      "step": 2097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957240309726348,
+      "loss": 0.0299,
+      "step": 2098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957194341436261,
+      "loss": 0.0063,
+      "step": 2099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957148348808277,
+      "loss": 0.0449,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029571023318431655,
+      "loss": 0.0356,
+      "step": 2101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957056290541695,
+      "loss": 0.045,
+      "step": 2102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029570102249046337,
+      "loss": 0.0121,
+      "step": 2103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956964134932751,
+      "loss": 0.0348,
+      "step": 2104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029569180206268173,
+      "loss": 0.0193,
+      "step": 2105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956871881987603,
+      "loss": 0.0158,
+      "step": 2106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029568257190158774,
+      "loss": 0.0337,
+      "step": 2107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029567795317124126,
+      "loss": 0.014,
+      "step": 2108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295673332007798,
+      "loss": 0.0161,
+      "step": 2109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956687084113351,
+      "loss": 0.0175,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956640823819298,
+      "loss": 0.0177,
+      "step": 2111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956594539196594,
+      "loss": 0.0429,
+      "step": 2112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029565482302460107,
+      "loss": 0.0177,
+      "step": 2113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029565018969683236,
+      "loss": 0.0075,
+      "step": 2114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956455539364305,
+      "loss": 0.0345,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295640915743473,
+      "loss": 0.0225,
+      "step": 2116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029563627511803724,
+      "loss": 0.0254,
+      "step": 2117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029563163206020076,
+      "loss": 0.0188,
+      "step": 2118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956269865700412,
+      "loss": 0.0308,
+      "step": 2119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295622338647636,
+      "loss": 0.0167,
+      "step": 2120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956176882930629,
+      "loss": 0.0374,
+      "step": 2121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029561303550639954,
+      "loss": 0.0248,
+      "step": 2122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956083802877236,
+      "loss": 0.0087,
+      "step": 2123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956037226371129,
+      "loss": 0.029,
+      "step": 2124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955990625546451,
+      "loss": 0.0186,
+      "step": 2125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955944000403982,
+      "loss": 0.0154,
+      "step": 2126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029558973509445,
+      "loss": 0.0346,
+      "step": 2127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029558506771687834,
+      "loss": 0.0339,
+      "step": 2128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029558039790776125,
+      "loss": 0.024,
+      "step": 2129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029557572566717675,
+      "loss": 0.0171,
+      "step": 2130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955710509952028,
+      "loss": 0.0292,
+      "step": 2131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029556637389191757,
+      "loss": 0.029,
+      "step": 2132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029556169435739906,
+      "loss": 0.0461,
+      "step": 2133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955570123917255,
+      "loss": 0.0444,
+      "step": 2134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029555232799497507,
+      "loss": 0.034,
+      "step": 2135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029554764116722604,
+      "loss": 0.0108,
+      "step": 2136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955429519085567,
+      "loss": 0.0411,
+      "step": 2137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029553826021904523,
+      "loss": 0.0079,
+      "step": 2138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029553356609877013,
+      "loss": 0.0169,
+      "step": 2139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029552886954780977,
+      "loss": 0.0192,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955241705662426,
+      "loss": 0.0573,
+      "step": 2141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029551946915414706,
+      "loss": 0.0305,
+      "step": 2142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955147653116017,
+      "loss": 0.015,
+      "step": 2143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955100590386851,
+      "loss": 0.0111,
+      "step": 2144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955053503354758,
+      "loss": 0.0206,
+      "step": 2145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029550063920205253,
+      "loss": 0.0165,
+      "step": 2146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029549592563849397,
+      "loss": 0.0134,
+      "step": 2147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029549120964487876,
+      "loss": 0.0148,
+      "step": 2148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954864912212857,
+      "loss": 0.0392,
+      "step": 2149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029548177036779364,
+      "loss": 0.0329,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954770470844814,
+      "loss": 0.0764,
+      "step": 2151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954723213714278,
+      "loss": 0.0562,
+      "step": 2152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954675932287119,
+      "loss": 0.0683,
+      "step": 2153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954628626564126,
+      "loss": 0.0339,
+      "step": 2154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954581296546089,
+      "loss": 0.0406,
+      "step": 2155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029545339422337984,
+      "loss": 0.0319,
+      "step": 2156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029544865636280457,
+      "loss": 0.0158,
+      "step": 2157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029544391607296217,
+      "loss": 0.0125,
+      "step": 2158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954391733539319,
+      "loss": 0.0257,
+      "step": 2159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954344282057928,
+      "loss": 0.0452,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029542968062862426,
+      "loss": 0.0417,
+      "step": 2161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954249306225055,
+      "loss": 0.0129,
+      "step": 2162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029542017818751597,
+      "loss": 0.028,
+      "step": 2163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954154233237349,
+      "loss": 0.0151,
+      "step": 2164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954106660312418,
+      "loss": 0.0503,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029540590631011603,
+      "loss": 0.0217,
+      "step": 2166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954011441604372,
+      "loss": 0.014,
+      "step": 2167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953963795822848,
+      "loss": 0.0367,
+      "step": 2168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029539161257573844,
+      "loss": 0.0109,
+      "step": 2169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953868431408776,
+      "loss": 0.0456,
+      "step": 2170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953820712777822,
+      "loss": 0.0408,
+      "step": 2171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029537729698653165,
+      "loss": 0.0395,
+      "step": 2172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029537252026720585,
+      "loss": 0.0528,
+      "step": 2173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953677411198846,
+      "loss": 0.0444,
+      "step": 2174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953629595446476,
+      "loss": 0.0261,
+      "step": 2175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029535817554157483,
+      "loss": 0.0278,
+      "step": 2176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953533891107461,
+      "loss": 0.0175,
+      "step": 2177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029534860025224143,
+      "loss": 0.008,
+      "step": 2178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953438089661408,
+      "loss": 0.017,
+      "step": 2179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953390152525242,
+      "loss": 0.043,
+      "step": 2180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029533421911147166,
+      "loss": 0.0057,
+      "step": 2181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029532942054306336,
+      "loss": 0.0191,
+      "step": 2182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953246195473794,
+      "loss": 0.0177,
+      "step": 2183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029531981612449995,
+      "loss": 0.0122,
+      "step": 2184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029531501027450525,
+      "loss": 0.0368,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002953102019974756,
+      "loss": 0.0185,
+      "step": 2186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002953053912934913,
+      "loss": 0.0243,
+      "step": 2187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029530057816263265,
+      "loss": 0.0276,
+      "step": 2188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952957626049801,
+      "loss": 0.0237,
+      "step": 2189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000295290944620614,
+      "loss": 0.0162,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029528612420961494,
+      "loss": 0.0319,
+      "step": 2191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952813013720633,
+      "loss": 0.0214,
+      "step": 2192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029527647610803974,
+      "loss": 0.0138,
+      "step": 2193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029527164841762477,
+      "loss": 0.0332,
+      "step": 2194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000295266818300899,
+      "loss": 0.0187,
+      "step": 2195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029526198575794327,
+      "loss": 0.0845,
+      "step": 2196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952571507888381,
+      "loss": 0.0432,
+      "step": 2197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029525231339366434,
+      "loss": 0.0449,
+      "step": 2198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952474735725028,
+      "loss": 0.035,
+      "step": 2199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952426313254342,
+      "loss": 0.0548,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952377866525395,
+      "loss": 0.0222,
+      "step": 2201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029523293955389967,
+      "loss": 0.013,
+      "step": 2202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952280900295955,
+      "loss": 0.0213,
+      "step": 2203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952232380797082,
+      "loss": 0.0198,
+      "step": 2204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952183837043187,
+      "loss": 0.0497,
+      "step": 2205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000295213526903508,
+      "loss": 0.0228,
+      "step": 2206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952086676773573,
+      "loss": 0.0453,
+      "step": 2207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029520380602594777,
+      "loss": 0.0263,
+      "step": 2208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029519894194936057,
+      "loss": 0.023,
+      "step": 2209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029519407544767695,
+      "loss": 0.0454,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951892065209782,
+      "loss": 0.0354,
+      "step": 2211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029518433516934566,
+      "loss": 0.0296,
+      "step": 2212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029517946139286067,
+      "loss": 0.0101,
+      "step": 2213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029517458519160464,
+      "loss": 0.0036,
+      "step": 2214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000295169706565659,
+      "loss": 0.0208,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951648255151052,
+      "loss": 0.021,
+      "step": 2216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029515994204002484,
+      "loss": 0.029,
+      "step": 2217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951550561404994,
+      "loss": 0.017,
+      "step": 2218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029515016781661053,
+      "loss": 0.0594,
+      "step": 2219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951452770684398,
+      "loss": 0.0469,
+      "step": 2220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029514038389606906,
+      "loss": 0.0293,
+      "step": 2221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951354882995799,
+      "loss": 0.0216,
+      "step": 2222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029513059027905415,
+      "loss": 0.0337,
+      "step": 2223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029512568983457356,
+      "loss": 0.0193,
+      "step": 2224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029512078696622006,
+      "loss": 0.0233,
+      "step": 2225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951158816740754,
+      "loss": 0.0731,
+      "step": 2226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951109739582216,
+      "loss": 0.0086,
+      "step": 2227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029510606381874066,
+      "loss": 0.0186,
+      "step": 2228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029510115125571453,
+      "loss": 0.0266,
+      "step": 2229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029509623626922533,
+      "loss": 0.0404,
+      "step": 2230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029509131885935503,
+      "loss": 0.0094,
+      "step": 2231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002950863990261858,
+      "loss": 0.0142,
+      "step": 2232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002950814767697999,
+      "loss": 0.0551,
+      "step": 2233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029507655209027944,
+      "loss": 0.0184,
+      "step": 2234
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029507162498770676,
+      "loss": 0.0174,
+      "step": 2235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029506669546216406,
+      "loss": 0.0276,
+      "step": 2236
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950617635137337,
+      "loss": 0.0064,
+      "step": 2237
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029505682914249807,
+      "loss": 0.0297,
+      "step": 2238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950518923485396,
+      "loss": 0.0126,
+      "step": 2239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029504695313194076,
+      "loss": 0.0099,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950420114927839,
+      "loss": 0.0146,
+      "step": 2241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029503706743115175,
+      "loss": 0.0223,
+      "step": 2242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950321209471268,
+      "loss": 0.0687,
+      "step": 2243
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029502717204079156,
+      "loss": 0.0069,
+      "step": 2244
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950222207122289,
+      "loss": 0.0134,
+      "step": 2245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029501726696152137,
+      "loss": 0.033,
+      "step": 2246
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029501231078875176,
+      "loss": 0.0228,
+      "step": 2247
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950073521940028,
+      "loss": 0.0152,
+      "step": 2248
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029500239117735737,
+      "loss": 0.0266,
+      "step": 2249
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029499742773889825,
+      "loss": 0.0236,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949924618787084,
+      "loss": 0.01,
+      "step": 2251
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029498749359687076,
+      "loss": 0.0084,
+      "step": 2252
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949825228934683,
+      "loss": 0.0487,
+      "step": 2253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000294977549768584,
+      "loss": 0.0169,
+      "step": 2254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000294972574222301,
+      "loss": 0.0206,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029496759625470233,
+      "loss": 0.0409,
+      "step": 2256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949626158658711,
+      "loss": 0.0093,
+      "step": 2257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029495763305589063,
+      "loss": 0.0397,
+      "step": 2258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000294952647824844,
+      "loss": 0.0335,
+      "step": 2259
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949476601728146,
+      "loss": 0.0048,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029494267009988565,
+      "loss": 0.0841,
+      "step": 2261
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949376776061405,
+      "loss": 0.0212,
+      "step": 2262
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029493268269166256,
+      "loss": 0.0248,
+      "step": 2263
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949276853565352,
+      "loss": 0.049,
+      "step": 2264
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949226856008419,
+      "loss": 0.0337,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029491768342466626,
+      "loss": 0.0515,
+      "step": 2266
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029491267882809176,
+      "loss": 0.013,
+      "step": 2267
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949076718112019,
+      "loss": 0.0187,
+      "step": 2268
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949026623740805,
+      "loss": 0.0206,
+      "step": 2269
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000294897650516811,
+      "loss": 0.0168,
+      "step": 2270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029489263623947733,
+      "loss": 0.0146,
+      "step": 2271
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029488761954216304,
+      "loss": 0.0238,
+      "step": 2272
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029488260042495205,
+      "loss": 0.021,
+      "step": 2273
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948775788879282,
+      "loss": 0.0136,
+      "step": 2274
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029487255493117524,
+      "loss": 0.0307,
+      "step": 2275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029486752855477713,
+      "loss": 0.0239,
+      "step": 2276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948624997588179,
+      "loss": 0.0076,
+      "step": 2277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029485746854338143,
+      "loss": 0.0333,
+      "step": 2278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948524349085518,
+      "loss": 0.0234,
+      "step": 2279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948473988544131,
+      "loss": 0.0624,
+      "step": 2280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029484236038104935,
+      "loss": 0.028,
+      "step": 2281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948373194885449,
+      "loss": 0.0108,
+      "step": 2282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948322761769837,
+      "loss": 0.0279,
+      "step": 2283
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029482723044645017,
+      "loss": 0.0147,
+      "step": 2284
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029482218229702847,
+      "loss": 0.0271,
+      "step": 2285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029481713172880295,
+      "loss": 0.0327,
+      "step": 2286
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294812078741858,
+      "loss": 0.0561,
+      "step": 2287
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029480702333627796,
+      "loss": 0.0366,
+      "step": 2288
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002948019655121473,
+      "loss": 0.0159,
+      "step": 2289
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029479690526955046,
+      "loss": 0.0222,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029479184260857203,
+      "loss": 0.0183,
+      "step": 2291
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947867775292965,
+      "loss": 0.0206,
+      "step": 2292
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029478171003180845,
+      "loss": 0.0402,
+      "step": 2293
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029477664011619254,
+      "loss": 0.0371,
+      "step": 2294
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947715677825335,
+      "loss": 0.0167,
+      "step": 2295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029476649303091597,
+      "loss": 0.0295,
+      "step": 2296
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947614158614247,
+      "loss": 0.0186,
+      "step": 2297
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029475633627414457,
+      "loss": 0.0143,
+      "step": 2298
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947512542691604,
+      "loss": 0.014,
+      "step": 2299
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029474616984655704,
+      "loss": 0.019,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029474108300641945,
+      "loss": 0.0103,
+      "step": 2301
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947359937488325,
+      "loss": 0.0524,
+      "step": 2302
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947309020738812,
+      "loss": 0.0183,
+      "step": 2303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947258079816507,
+      "loss": 0.0267,
+      "step": 2304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294720711472226,
+      "loss": 0.0326,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947156125456923,
+      "loss": 0.0334,
+      "step": 2306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029471051120213466,
+      "loss": 0.0246,
+      "step": 2307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947054074416383,
+      "loss": 0.012,
+      "step": 2308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947003012642885,
+      "loss": 0.0276,
+      "step": 2309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946951926701705,
+      "loss": 0.0266,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946900816593697,
+      "loss": 0.0318,
+      "step": 2311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946849682319714,
+      "loss": 0.0183,
+      "step": 2312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294679852388061,
+      "loss": 0.0116,
+      "step": 2313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294674734127724,
+      "loss": 0.0108,
+      "step": 2314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946696134510458,
+      "loss": 0.021,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029466449035811195,
+      "loss": 0.0152,
+      "step": 2316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946593648490081,
+      "loss": 0.0464,
+      "step": 2317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029465423692381974,
+      "loss": 0.0185,
+      "step": 2318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946491065826326,
+      "loss": 0.0359,
+      "step": 2319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029464397382553234,
+      "loss": 0.0137,
+      "step": 2320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946388386526047,
+      "loss": 0.0229,
+      "step": 2321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946337010639354,
+      "loss": 0.0162,
+      "step": 2322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946285610596103,
+      "loss": 0.0064,
+      "step": 2323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946234186397152,
+      "loss": 0.0144,
+      "step": 2324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029461827380433605,
+      "loss": 0.0366,
+      "step": 2325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946131265535587,
+      "loss": 0.0126,
+      "step": 2326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029460797688746914,
+      "loss": 0.0261,
+      "step": 2327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946028248061535,
+      "loss": 0.0506,
+      "step": 2328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002945976703096976,
+      "loss": 0.0401,
+      "step": 2329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002945925133981877,
+      "loss": 0.0052,
+      "step": 2330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029458735407170996,
+      "loss": 0.0189,
+      "step": 2331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002945821923303504,
+      "loss": 0.038,
+      "step": 2332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945770281741953,
+      "loss": 0.0121,
+      "step": 2333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000294571861603331,
+      "loss": 0.0377,
+      "step": 2334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945666926178437,
+      "loss": 0.0403,
+      "step": 2335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945615212178197,
+      "loss": 0.0169,
+      "step": 2336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945563474033454,
+      "loss": 0.0277,
+      "step": 2337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945511711745073,
+      "loss": 0.0478,
+      "step": 2338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945459925313917,
+      "loss": 0.013,
+      "step": 2339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029454081147408523,
+      "loss": 0.0298,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029453562800267434,
+      "loss": 0.0411,
+      "step": 2341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029453044211724563,
+      "loss": 0.0449,
+      "step": 2342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945252538178858,
+      "loss": 0.017,
+      "step": 2343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945200631046813,
+      "loss": 0.0262,
+      "step": 2344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000294514869977719,
+      "loss": 0.0231,
+      "step": 2345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945096744370856,
+      "loss": 0.0149,
+      "step": 2346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945044764828678,
+      "loss": 0.0166,
+      "step": 2347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029449927611515254,
+      "loss": 0.0178,
+      "step": 2348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944940733340266,
+      "loss": 0.025,
+      "step": 2349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944888681395768,
+      "loss": 0.0288,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029448366053189026,
+      "loss": 0.0292,
+      "step": 2351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944784505110538,
+      "loss": 0.0263,
+      "step": 2352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944732380771545,
+      "loss": 0.022,
+      "step": 2353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944680232302794,
+      "loss": 0.0301,
+      "step": 2354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029446280597051564,
+      "loss": 0.0249,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944575862979503,
+      "loss": 0.0181,
+      "step": 2356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944523642126706,
+      "loss": 0.0144,
+      "step": 2357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944471397147638,
+      "loss": 0.0107,
+      "step": 2358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000294441912804317,
+      "loss": 0.0268,
+      "step": 2359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029443668348141765,
+      "loss": 0.0072,
+      "step": 2360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029443145174615307,
+      "loss": 0.0099,
+      "step": 2361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029442621759861055,
+      "loss": 0.0252,
+      "step": 2362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944209810388776,
+      "loss": 0.021,
+      "step": 2363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944157420670417,
+      "loss": 0.0365,
+      "step": 2364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944105006831902,
+      "loss": 0.0281,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944052568874109,
+      "loss": 0.0118,
+      "step": 2366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944000106797911,
+      "loss": 0.0327,
+      "step": 2367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943947620604186,
+      "loss": 0.004,
+      "step": 2368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029438951102938097,
+      "loss": 0.0248,
+      "step": 2369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029438425758676596,
+      "loss": 0.0196,
+      "step": 2370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943790017326613,
+      "loss": 0.0315,
+      "step": 2371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943737434671548,
+      "loss": 0.0106,
+      "step": 2372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943684827903343,
+      "loss": 0.0163,
+      "step": 2373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943632197022875,
+      "loss": 0.1069,
+      "step": 2374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943579542031025,
+      "loss": 0.0514,
+      "step": 2375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943526862928672,
+      "loss": 0.0102,
+      "step": 2376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943474159716696,
+      "loss": 0.0495,
+      "step": 2377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029434214323959755,
+      "loss": 0.0246,
+      "step": 2378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029433686809673926,
+      "loss": 0.022,
+      "step": 2379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943315905431829,
+      "loss": 0.0425,
+      "step": 2380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029432631057901647,
+      "loss": 0.0356,
+      "step": 2381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002943210282043282,
+      "loss": 0.007,
+      "step": 2382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002943157434192064,
+      "loss": 0.014,
+      "step": 2383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002943104562237393,
+      "loss": 0.0228,
+      "step": 2384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029430516661801503,
+      "loss": 0.0188,
+      "step": 2385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942998746021222,
+      "loss": 0.0376,
+      "step": 2386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000294294580176149,
+      "loss": 0.0114,
+      "step": 2387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029428928334018404,
+      "loss": 0.0198,
+      "step": 2388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942839840943156,
+      "loss": 0.0128,
+      "step": 2389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942786824386323,
+      "loss": 0.0318,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942733783732226,
+      "loss": 0.0116,
+      "step": 2391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942680718981752,
+      "loss": 0.0433,
+      "step": 2392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942627630135787,
+      "loss": 0.0345,
+      "step": 2393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029425745171952173,
+      "loss": 0.0433,
+      "step": 2394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029425213801609297,
+      "loss": 0.0093,
+      "step": 2395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029424682190338125,
+      "loss": 0.061,
+      "step": 2396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029424150338147533,
+      "loss": 0.0303,
+      "step": 2397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000294236182450464,
+      "loss": 0.0067,
+      "step": 2398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029423085911043615,
+      "loss": 0.0138,
+      "step": 2399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942255333614807,
+      "loss": 0.0412,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942202052036866,
+      "loss": 0.0068,
+      "step": 2401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942148746371428,
+      "loss": 0.0132,
+      "step": 2402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942095416619384,
+      "loss": 0.0911,
+      "step": 2403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029420420627816247,
+      "loss": 0.0233,
+      "step": 2404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941988684859041,
+      "loss": 0.0526,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941935282852524,
+      "loss": 0.0133,
+      "step": 2406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941881856762966,
+      "loss": 0.0275,
+      "step": 2407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029418284065912595,
+      "loss": 0.0274,
+      "step": 2408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029417749323382966,
+      "loss": 0.0032,
+      "step": 2409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029417214340049703,
+      "loss": 0.0313,
+      "step": 2410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941667911592175,
+      "loss": 0.0182,
+      "step": 2411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941614365100805,
+      "loss": 0.0114,
+      "step": 2412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941560794531753,
+      "loss": 0.0043,
+      "step": 2413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941507199885915,
+      "loss": 0.0141,
+      "step": 2414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029414535811641857,
+      "loss": 0.0165,
+      "step": 2415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029413999383674604,
+      "loss": 0.0295,
+      "step": 2416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029413462714966357,
+      "loss": 0.0215,
+      "step": 2417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941292580552607,
+      "loss": 0.006,
+      "step": 2418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941238865536272,
+      "loss": 0.0404,
+      "step": 2419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029411851264485273,
+      "loss": 0.0374,
+      "step": 2420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029411313632902704,
+      "loss": 0.0629,
+      "step": 2421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029410775760623994,
+      "loss": 0.0097,
+      "step": 2422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029410237647658127,
+      "loss": 0.0092,
+      "step": 2423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029409699294014095,
+      "loss": 0.0278,
+      "step": 2424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029409160699700877,
+      "loss": 0.0176,
+      "step": 2425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002940862186472748,
+      "loss": 0.0398,
+      "step": 2426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000294080827891029,
+      "loss": 0.0164,
+      "step": 2427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002940754347283614,
+      "loss": 0.0339,
+      "step": 2428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029407003915936207,
+      "loss": 0.026,
+      "step": 2429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002940646411841211,
+      "loss": 0.0155,
+      "step": 2430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002940592408027288,
+      "loss": 0.0341,
+      "step": 2431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029405383801527513,
+      "loss": 0.0318,
+      "step": 2432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029404843282185045,
+      "loss": 0.0283,
+      "step": 2433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940430252225451,
+      "loss": 0.0163,
+      "step": 2434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940376152174493,
+      "loss": 0.0316,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029403220280665337,
+      "loss": 0.0121,
+      "step": 2436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029402678799024783,
+      "loss": 0.0208,
+      "step": 2437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940213707683231,
+      "loss": 0.0406,
+      "step": 2438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029401595114096954,
+      "loss": 0.0269,
+      "step": 2439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940105291082778,
+      "loss": 0.0231,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940051046703384,
+      "loss": 0.0529,
+      "step": 2441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029399967782724185,
+      "loss": 0.0257,
+      "step": 2442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029399424857907893,
+      "loss": 0.0195,
+      "step": 2443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029398881692594016,
+      "loss": 0.0443,
+      "step": 2444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029398338286791646,
+      "loss": 0.0188,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029397794640509845,
+      "loss": 0.0145,
+      "step": 2446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029397250753757694,
+      "loss": 0.0137,
+      "step": 2447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029396706626544274,
+      "loss": 0.0217,
+      "step": 2448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939616225887868,
+      "loss": 0.0093,
+      "step": 2449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029395617650770006,
+      "loss": 0.014,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029395072802227345,
+      "loss": 0.0231,
+      "step": 2451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939452771325979,
+      "loss": 0.0066,
+      "step": 2452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939398238387645,
+      "loss": 0.0705,
+      "step": 2453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939343681408644,
+      "loss": 0.0357,
+      "step": 2454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939289100389886,
+      "loss": 0.0508,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939234495332283,
+      "loss": 0.016,
+      "step": 2456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939179866236748,
+      "loss": 0.0223,
+      "step": 2457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939125213104192,
+      "loss": 0.0182,
+      "step": 2458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029390705359355287,
+      "loss": 0.0079,
+      "step": 2459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939015834731671,
+      "loss": 0.0142,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938961109493532,
+      "loss": 0.0205,
+      "step": 2461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029389063602220267,
+      "loss": 0.0111,
+      "step": 2462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938851586918069,
+      "loss": 0.0168,
+      "step": 2463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938796789582574,
+      "loss": 0.0253,
+      "step": 2464
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938741968216456,
+      "loss": 0.0122,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029386871228206316,
+      "loss": 0.0286,
+      "step": 2466
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938632253396017,
+      "loss": 0.0149,
+      "step": 2467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029385773599435277,
+      "loss": 0.0246,
+      "step": 2468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938522442464081,
+      "loss": 0.0124,
+      "step": 2469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029384675009585945,
+      "loss": 0.0213,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938412535427985,
+      "loss": 0.0215,
+      "step": 2471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938357545873171,
+      "loss": 0.0134,
+      "step": 2472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938302532295071,
+      "loss": 0.0449,
+      "step": 2473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938247494694604,
+      "loss": 0.0236,
+      "step": 2474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938192433072688,
+      "loss": 0.0225,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938137347430245,
+      "loss": 0.0202,
+      "step": 2476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029380822377681924,
+      "loss": 0.0198,
+      "step": 2477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029380271040874525,
+      "loss": 0.02,
+      "step": 2478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002937971946388945,
+      "loss": 0.0283,
+      "step": 2479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029379167646735924,
+      "loss": 0.0176,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937861558942315,
+      "loss": 0.0244,
+      "step": 2481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029378063291960355,
+      "loss": 0.0252,
+      "step": 2482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029377510754356766,
+      "loss": 0.0133,
+      "step": 2483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029376957976621603,
+      "loss": 0.0419,
+      "step": 2484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029376404958764103,
+      "loss": 0.0332,
+      "step": 2485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937585170079351,
+      "loss": 0.0038,
+      "step": 2486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937529820271905,
+      "loss": 0.0253,
+      "step": 2487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029374744464549975,
+      "loss": 0.0162,
+      "step": 2488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937419048629554,
+      "loss": 0.0249,
+      "step": 2489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029373636267964983,
+      "loss": 0.0426,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937308180956757,
+      "loss": 0.0198,
+      "step": 2491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937252711111256,
+      "loss": 0.0438,
+      "step": 2492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937197217260922,
+      "loss": 0.0397,
+      "step": 2493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029371416994066813,
+      "loss": 0.0173,
+      "step": 2494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029370861575494616,
+      "loss": 0.0266,
+      "step": 2495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000293703059169019,
+      "loss": 0.0191,
+      "step": 2496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029369750018297955,
+      "loss": 0.0137,
+      "step": 2497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936919387969205,
+      "loss": 0.0231,
+      "step": 2498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029368637501093494,
+      "loss": 0.0174,
+      "step": 2499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029368080882511566,
+      "loss": 0.035,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029367524023955565,
+      "loss": 0.0097,
+      "step": 2501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029366966925434794,
+      "loss": 0.0375,
+      "step": 2502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936640958695855,
+      "loss": 0.0385,
+      "step": 2503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936585200853615,
+      "loss": 0.0147,
+      "step": 2504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936529419017691,
+      "loss": 0.055,
+      "step": 2505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936473613189013,
+      "loss": 0.0334,
+      "step": 2506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029364177833685154,
+      "loss": 0.0626,
+      "step": 2507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936361929557129,
+      "loss": 0.0245,
+      "step": 2508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936306051755786,
+      "loss": 0.0171,
+      "step": 2509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029362501499654214,
+      "loss": 0.0053,
+      "step": 2510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936194224186968,
+      "loss": 0.0335,
+      "step": 2511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029361382744213604,
+      "loss": 0.0102,
+      "step": 2512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936082300669533,
+      "loss": 0.0183,
+      "step": 2513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000293602630293242,
+      "loss": 0.0263,
+      "step": 2514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029359702812109566,
+      "loss": 0.021,
+      "step": 2515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029359142355060795,
+      "loss": 0.0239,
+      "step": 2516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935858165818724,
+      "loss": 0.0237,
+      "step": 2517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935802072149827,
+      "loss": 0.0327,
+      "step": 2518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029357459545003244,
+      "loss": 0.0646,
+      "step": 2519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935689812871155,
+      "loss": 0.0391,
+      "step": 2520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935633647263255,
+      "loss": 0.0395,
+      "step": 2521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935577457677564,
+      "loss": 0.0223,
+      "step": 2522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935521244115019,
+      "loss": 0.0391,
+      "step": 2523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000293546500657656,
+      "loss": 0.0386,
+      "step": 2524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029354087450631254,
+      "loss": 0.0354,
+      "step": 2525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935352459575655,
+      "loss": 0.0163,
+      "step": 2526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029352961501150895,
+      "loss": 0.0106,
+      "step": 2527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935239816682369,
+      "loss": 0.0342,
+      "step": 2528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935183459278434,
+      "loss": 0.0161,
+      "step": 2529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002935127077904226,
+      "loss": 0.0287,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002935070672560687,
+      "loss": 0.0133,
+      "step": 2531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002935014243248759,
+      "loss": 0.0117,
+      "step": 2532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934957789969384,
+      "loss": 0.0256,
+      "step": 2533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029349013127235056,
+      "loss": 0.0259,
+      "step": 2534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029348448115120666,
+      "loss": 0.0128,
+      "step": 2535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934788286336011,
+      "loss": 0.0104,
+      "step": 2536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029347317371962824,
+      "loss": 0.0047,
+      "step": 2537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934675164093826,
+      "loss": 0.0266,
+      "step": 2538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934618567029586,
+      "loss": 0.0197,
+      "step": 2539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029345619460045076,
+      "loss": 0.0114,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934505301019537,
+      "loss": 0.0201,
+      "step": 2541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293444863207562,
+      "loss": 0.0183,
+      "step": 2542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934391939173703,
+      "loss": 0.0308,
+      "step": 2543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934335222314733,
+      "loss": 0.0668,
+      "step": 2544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029342784814996575,
+      "loss": 0.0255,
+      "step": 2545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029342217167294235,
+      "loss": 0.0104,
+      "step": 2546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293416492800498,
+      "loss": 0.0028,
+      "step": 2547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029341081153272754,
+      "loss": 0.0246,
+      "step": 2548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029340512786972577,
+      "loss": 0.0212,
+      "step": 2549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029339944181158765,
+      "loss": 0.0243,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933937533584082,
+      "loss": 0.0144,
+      "step": 2551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933880625102824,
+      "loss": 0.0121,
+      "step": 2552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029338236926730523,
+      "loss": 0.0261,
+      "step": 2553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933766736295719,
+      "loss": 0.026,
+      "step": 2554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029337097559717747,
+      "loss": 0.0532,
+      "step": 2555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933652751702171,
+      "loss": 0.0266,
+      "step": 2556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293359572348786,
+      "loss": 0.0357,
+      "step": 2557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029335386713297946,
+      "loss": 0.0107,
+      "step": 2558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933481595228928,
+      "loss": 0.0318,
+      "step": 2559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933424495186212,
+      "loss": 0.0411,
+      "step": 2560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933367371202601,
+      "loss": 0.0203,
+      "step": 2561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029333102232790496,
+      "loss": 0.0109,
+      "step": 2562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029332530514165125,
+      "loss": 0.036,
+      "step": 2563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029331958556159433,
+      "loss": 0.0464,
+      "step": 2564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933138635878298,
+      "loss": 0.0209,
+      "step": 2565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933081392204532,
+      "loss": 0.0103,
+      "step": 2566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029330241245956016,
+      "loss": 0.0333,
+      "step": 2567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932966833052464,
+      "loss": 0.0395,
+      "step": 2568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932909517576075,
+      "loss": 0.022,
+      "step": 2569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932852178167392,
+      "loss": 0.0239,
+      "step": 2570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029327948148273733,
+      "loss": 0.0142,
+      "step": 2571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029327374275569764,
+      "loss": 0.0213,
+      "step": 2572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293268001635716,
+      "loss": 0.0391,
+      "step": 2573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029326225812288827,
+      "loss": 0.0142,
+      "step": 2574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029325651221731044,
+      "loss": 0.0135,
+      "step": 2575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029325076391907844,
+      "loss": 0.0127,
+      "step": 2576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932450132282883,
+      "loss": 0.0211,
+      "step": 2577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293239260145036,
+      "loss": 0.003,
+      "step": 2578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029323350466941764,
+      "loss": 0.014,
+      "step": 2579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029322774680152943,
+      "loss": 0.0074,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029322198654146747,
+      "loss": 0.0154,
+      "step": 2581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000293216223889328,
+      "loss": 0.0143,
+      "step": 2582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029321045884520724,
+      "loss": 0.0151,
+      "step": 2583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002932046914092015,
+      "loss": 0.0487,
+      "step": 2584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931989215814071,
+      "loss": 0.0214,
+      "step": 2585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029319314936192037,
+      "loss": 0.0121,
+      "step": 2586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931873747508378,
+      "loss": 0.0144,
+      "step": 2587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931815977482558,
+      "loss": 0.0081,
+      "step": 2588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029317581835427075,
+      "loss": 0.0283,
+      "step": 2589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029317003656897933,
+      "loss": 0.0101,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000293164252392478,
+      "loss": 0.01,
+      "step": 2591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029315846582486345,
+      "loss": 0.0096,
+      "step": 2592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029315267686623236,
+      "loss": 0.0264,
+      "step": 2593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029314688551668124,
+      "loss": 0.0223,
+      "step": 2594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029314109177630695,
+      "loss": 0.0918,
+      "step": 2595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029313529564520626,
+      "loss": 0.0084,
+      "step": 2596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931294971234759,
+      "loss": 0.0377,
+      "step": 2597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931236962112128,
+      "loss": 0.0347,
+      "step": 2598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029311789290851375,
+      "loss": 0.0228,
+      "step": 2599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029311208721547583,
+      "loss": 0.0547,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931062791321959,
+      "loss": 0.0522,
+      "step": 2601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931004686587709,
+      "loss": 0.0732,
+      "step": 2602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000293094655795298,
+      "loss": 0.0132,
+      "step": 2603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930888405418742,
+      "loss": 0.0368,
+      "step": 2604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029308302289859673,
+      "loss": 0.013,
+      "step": 2605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930772028655626,
+      "loss": 0.0199,
+      "step": 2606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029307138044286916,
+      "loss": 0.0423,
+      "step": 2607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029306555563061364,
+      "loss": 0.0293,
+      "step": 2608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930597284288932,
+      "loss": 0.0413,
+      "step": 2609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930538988378053,
+      "loss": 0.067,
+      "step": 2610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930480668574472,
+      "loss": 0.0457,
+      "step": 2611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930422324879165,
+      "loss": 0.0185,
+      "step": 2612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029303639572931037,
+      "loss": 0.0198,
+      "step": 2613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029303055658172646,
+      "loss": 0.0067,
+      "step": 2614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930247150452623,
+      "loss": 0.0196,
+      "step": 2615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930188711200154,
+      "loss": 0.0339,
+      "step": 2616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029301302480608336,
+      "loss": 0.0298,
+      "step": 2617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930071761035639,
+      "loss": 0.0216,
+      "step": 2618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930013250125546,
+      "loss": 0.0204,
+      "step": 2619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929954715331533,
+      "loss": 0.0324,
+      "step": 2620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029298961566545764,
+      "loss": 0.0385,
+      "step": 2621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029298375740956554,
+      "loss": 0.0223,
+      "step": 2622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029297789676557476,
+      "loss": 0.0153,
+      "step": 2623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929720337335832,
+      "loss": 0.0147,
+      "step": 2624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029296616831368886,
+      "loss": 0.0341,
+      "step": 2625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929603005059896,
+      "loss": 0.0315,
+      "step": 2626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029295443031058344,
+      "loss": 0.0037,
+      "step": 2627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029294855772756846,
+      "loss": 0.029,
+      "step": 2628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029294268275704277,
+      "loss": 0.047,
+      "step": 2629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929368053991044,
+      "loss": 0.0223,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029293092565385164,
+      "loss": 0.0549,
+      "step": 2631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029292504352138255,
+      "loss": 0.0674,
+      "step": 2632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929191590017955,
+      "loss": 0.0255,
+      "step": 2633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029291327209518867,
+      "loss": 0.0537,
+      "step": 2634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029290738280166045,
+      "loss": 0.0431,
+      "step": 2635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929014911213092,
+      "loss": 0.0777,
+      "step": 2636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029289559705423325,
+      "loss": 0.0265,
+      "step": 2637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928897006005311,
+      "loss": 0.0189,
+      "step": 2638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029288380176030124,
+      "loss": 0.0242,
+      "step": 2639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928779005336422,
+      "loss": 0.0405,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928719969206525,
+      "loss": 0.0193,
+      "step": 2641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928660909214308,
+      "loss": 0.0204,
+      "step": 2642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029286018253607565,
+      "loss": 0.0063,
+      "step": 2643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928542717646858,
+      "loss": 0.0225,
+      "step": 2644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029284835860735997,
+      "loss": 0.0171,
+      "step": 2645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928424430641969,
+      "loss": 0.0413,
+      "step": 2646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029283652513529537,
+      "loss": 0.0384,
+      "step": 2647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928306048207543,
+      "loss": 0.0101,
+      "step": 2648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029282468212067246,
+      "loss": 0.0042,
+      "step": 2649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029281875703514884,
+      "loss": 0.0201,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029281282956428244,
+      "loss": 0.0184,
+      "step": 2651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029280689970817217,
+      "loss": 0.0057,
+      "step": 2652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928009674669171,
+      "loss": 0.0647,
+      "step": 2653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927950328406163,
+      "loss": 0.0146,
+      "step": 2654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927890958293689,
+      "loss": 0.0187,
+      "step": 2655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927831564332741,
+      "loss": 0.0128,
+      "step": 2656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927772146524311,
+      "loss": 0.0168,
+      "step": 2657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927712704869391,
+      "loss": 0.01,
+      "step": 2658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029276532393689733,
+      "loss": 0.0675,
+      "step": 2659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927593750024052,
+      "loss": 0.0157,
+      "step": 2660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000292753423683562,
+      "loss": 0.0083,
+      "step": 2661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927474699804672,
+      "loss": 0.0369,
+      "step": 2662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029274151389322014,
+      "loss": 0.0158,
+      "step": 2663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927355554219204,
+      "loss": 0.0168,
+      "step": 2664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927295945666674,
+      "loss": 0.0194,
+      "step": 2665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927236313275608,
+      "loss": 0.0128,
+      "step": 2666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029271766570470014,
+      "loss": 0.0435,
+      "step": 2667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029271169769818506,
+      "loss": 0.0298,
+      "step": 2668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927057273081152,
+      "loss": 0.0132,
+      "step": 2669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926997545345904,
+      "loss": 0.0078,
+      "step": 2670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926937793777102,
+      "loss": 0.038,
+      "step": 2671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926878018375747,
+      "loss": 0.0073,
+      "step": 2672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029268182191428347,
+      "loss": 0.0197,
+      "step": 2673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926758396079365,
+      "loss": 0.047,
+      "step": 2674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926698549186337,
+      "loss": 0.0207,
+      "step": 2675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029266386784647494,
+      "loss": 0.0417,
+      "step": 2676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926578783915604,
+      "loss": 0.0103,
+      "step": 2677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029265188655398994,
+      "loss": 0.0266,
+      "step": 2678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926458923338637,
+      "loss": 0.0437,
+      "step": 2679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926398957312818,
+      "loss": 0.0682,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029263389674634443,
+      "loss": 0.0527,
+      "step": 2681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029262789537915174,
+      "loss": 0.0155,
+      "step": 2682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926218916298039,
+      "loss": 0.0384,
+      "step": 2683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926158854984013,
+      "loss": 0.0337,
+      "step": 2684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029260987698504425,
+      "loss": 0.0133,
+      "step": 2685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000292603866089833,
+      "loss": 0.0048,
+      "step": 2686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000292597852812868,
+      "loss": 0.0163,
+      "step": 2687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925918371542497,
+      "loss": 0.0237,
+      "step": 2688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029258581911407853,
+      "loss": 0.0366,
+      "step": 2689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029257979869245506,
+      "loss": 0.0425,
+      "step": 2690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925737758894798,
+      "loss": 0.0084,
+      "step": 2691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925677507052534,
+      "loss": 0.0349,
+      "step": 2692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029256172313987637,
+      "loss": 0.007,
+      "step": 2693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029255569319344953,
+      "loss": 0.0052,
+      "step": 2694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925496608660735,
+      "loss": 0.0267,
+      "step": 2695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925436261578491,
+      "loss": 0.0186,
+      "step": 2696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029253758906887695,
+      "loss": 0.0373,
+      "step": 2697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925315495992581,
+      "loss": 0.0072,
+      "step": 2698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925255077490933,
+      "loss": 0.0246,
+      "step": 2699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029251946351848345,
+      "loss": 0.0329,
+      "step": 2700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925134169075296,
+      "loss": 0.0197,
+      "step": 2701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029250736791633256,
+      "loss": 0.0333,
+      "step": 2702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925013165449936,
+      "loss": 0.0304,
+      "step": 2703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924952627936136,
+      "loss": 0.0024,
+      "step": 2704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924892066622937,
+      "loss": 0.0462,
+      "step": 2705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924831481511352,
+      "loss": 0.0181,
+      "step": 2706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924770872602391,
+      "loss": 0.0192,
+      "step": 2707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924710239897067,
+      "loss": 0.018,
+      "step": 2708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924649583396393,
+      "loss": 0.0304,
+      "step": 2709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924588903101381,
+      "loss": 0.0145,
+      "step": 2710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924528199013045,
+      "loss": 0.0094,
+      "step": 2711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029244674711324,
+      "loss": 0.0299,
+      "step": 2712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029244067194604594,
+      "loss": 0.0142,
+      "step": 2713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029243459439982374,
+      "loss": 0.0118,
+      "step": 2714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029242851447467495,
+      "loss": 0.0414,
+      "step": 2715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924224321707011,
+      "loss": 0.0181,
+      "step": 2716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029241634748800385,
+      "loss": 0.0304,
+      "step": 2717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924102604266847,
+      "loss": 0.0305,
+      "step": 2718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029240417098684547,
+      "loss": 0.0108,
+      "step": 2719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002923980791685877,
+      "loss": 0.0159,
+      "step": 2720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002923919849720132,
+      "loss": 0.0272,
+      "step": 2721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029238588839722383,
+      "loss": 0.0145,
+      "step": 2722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029237978944432133,
+      "loss": 0.021,
+      "step": 2723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002923736881134076,
+      "loss": 0.0385,
+      "step": 2724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002923675844045845,
+      "loss": 0.0102,
+      "step": 2725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292361478317954,
+      "loss": 0.033,
+      "step": 2726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029235536985361806,
+      "loss": 0.0144,
+      "step": 2727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029234925901167877,
+      "loss": 0.0283,
+      "step": 2728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923431457922382,
+      "loss": 0.0101,
+      "step": 2729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923370301953983,
+      "loss": 0.0139,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923309122212614,
+      "loss": 0.0072,
+      "step": 2731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029232479186992955,
+      "loss": 0.0103,
+      "step": 2732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292318669141505,
+      "loss": 0.0368,
+      "step": 2733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923125440360901,
+      "loss": 0.014,
+      "step": 2734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292306416553787,
+      "loss": 0.0302,
+      "step": 2735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029230028669469815,
+      "loss": 0.0507,
+      "step": 2736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029229415445892597,
+      "loss": 0.0446,
+      "step": 2737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922880198465727,
+      "loss": 0.0378,
+      "step": 2738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292281882857741,
+      "loss": 0.0243,
+      "step": 2739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029227574349253326,
+      "loss": 0.0254,
+      "step": 2740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292269601751052,
+      "loss": 0.0243,
+      "step": 2741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922634576333999,
+      "loss": 0.038,
+      "step": 2742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029225731113967943,
+      "loss": 0.0109,
+      "step": 2743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922511622699934,
+      "loss": 0.0098,
+      "step": 2744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029224501102444444,
+      "loss": 0.017,
+      "step": 2745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029223885740313526,
+      "loss": 0.0035,
+      "step": 2746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922327014061687,
+      "loss": 0.0198,
+      "step": 2747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029222654303364754,
+      "loss": 0.007,
+      "step": 2748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029222038228567457,
+      "loss": 0.0105,
+      "step": 2749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922142191623528,
+      "loss": 0.0137,
+      "step": 2750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922080536637852,
+      "loss": 0.017,
+      "step": 2751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029220188579007455,
+      "loss": 0.0686,
+      "step": 2752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921957155413241,
+      "loss": 0.0359,
+      "step": 2753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921895429176367,
+      "loss": 0.0351,
+      "step": 2754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921833679191155,
+      "loss": 0.0314,
+      "step": 2755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921771905458637,
+      "loss": 0.0164,
+      "step": 2756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921710107979845,
+      "loss": 0.0134,
+      "step": 2757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292164828675581,
+      "loss": 0.0193,
+      "step": 2758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921586441787565,
+      "loss": 0.0143,
+      "step": 2759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921524573076143,
+      "loss": 0.0054,
+      "step": 2760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029214626806225774,
+      "loss": 0.021,
+      "step": 2761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921400764427902,
+      "loss": 0.0171,
+      "step": 2762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921338824493151,
+      "loss": 0.0175,
+      "step": 2763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921276860819358,
+      "loss": 0.0448,
+      "step": 2764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029212148734075586,
+      "loss": 0.0141,
+      "step": 2765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029211528622587885,
+      "loss": 0.0118,
+      "step": 2766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921090827374083,
+      "loss": 0.0101,
+      "step": 2767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029210287687544777,
+      "loss": 0.0433,
+      "step": 2768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000292096668640101,
+      "loss": 0.0121,
+      "step": 2769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029209045803147164,
+      "loss": 0.0187,
+      "step": 2770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029208424504966336,
+      "loss": 0.0473,
+      "step": 2771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002920780296947801,
+      "loss": 0.0124,
+      "step": 2772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002920718119669254,
+      "loss": 0.0189,
+      "step": 2773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002920655918662034,
+      "loss": 0.0191,
+      "step": 2774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029205936939271775,
+      "loss": 0.0417,
+      "step": 2775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029205314454657254,
+      "loss": 0.0479,
+      "step": 2776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920469173278716,
+      "loss": 0.0239,
+      "step": 2777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920406877367191,
+      "loss": 0.0411,
+      "step": 2778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920344557732189,
+      "loss": 0.0077,
+      "step": 2779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029202822143747524,
+      "loss": 0.028,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920219847295922,
+      "loss": 0.0144,
+      "step": 2781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920157456496739,
+      "loss": 0.0268,
+      "step": 2782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920095041978245,
+      "loss": 0.0504,
+      "step": 2783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029200326037414845,
+      "loss": 0.0465,
+      "step": 2784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029199701417874987,
+      "loss": 0.0163,
+      "step": 2785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919907656117331,
+      "loss": 0.0221,
+      "step": 2786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029198451467320245,
+      "loss": 0.0393,
+      "step": 2787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919782613632625,
+      "loss": 0.0123,
+      "step": 2788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919720056820175,
+      "loss": 0.0217,
+      "step": 2789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291965747629572,
+      "loss": 0.0305,
+      "step": 2790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919594872060306,
+      "loss": 0.0116,
+      "step": 2791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919532244114978,
+      "loss": 0.0295,
+      "step": 2792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029194695924607814,
+      "loss": 0.0268,
+      "step": 2793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029194069170987636,
+      "loss": 0.0085,
+      "step": 2794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919344218029971,
+      "loss": 0.0198,
+      "step": 2795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029192814952554505,
+      "loss": 0.014,
+      "step": 2796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029192187487762504,
+      "loss": 0.0183,
+      "step": 2797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029191559785934174,
+      "loss": 0.0282,
+      "step": 2798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919093184708001,
+      "loss": 0.0159,
+      "step": 2799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291903036712105,
+      "loss": 0.0184,
+      "step": 2800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029189675258336134,
+      "loss": 0.0186,
+      "step": 2801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291890466084674,
+      "loss": 0.0224,
+      "step": 2802
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029188417721614813,
+      "loss": 0.0122,
+      "step": 2803
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029187788597788856,
+      "loss": 0.0251,
+      "step": 2804
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029187159237000057,
+      "loss": 0.0255,
+      "step": 2805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029186529639258917,
+      "loss": 0.0209,
+      "step": 2806
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918589980457595,
+      "loss": 0.0161,
+      "step": 2807
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918526973296168,
+      "loss": 0.026,
+      "step": 2808
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029184639424426626,
+      "loss": 0.0345,
+      "step": 2809
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029184008878981325,
+      "loss": 0.0161,
+      "step": 2810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291833780966363,
+      "loss": 0.0081,
+      "step": 2811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918274707740208,
+      "loss": 0.0081,
+      "step": 2812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918211582128922,
+      "loss": 0.0211,
+      "step": 2813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918148432830825,
+      "loss": 0.013,
+      "step": 2814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029180852598469727,
+      "loss": 0.032,
+      "step": 2815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029180220631784196,
+      "loss": 0.0332,
+      "step": 2816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029179588428262214,
+      "loss": 0.0163,
+      "step": 2817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917895598791434,
+      "loss": 0.0206,
+      "step": 2818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917832331075113,
+      "loss": 0.0362,
+      "step": 2819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917769039678316,
+      "loss": 0.0336,
+      "step": 2820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029177057246021005,
+      "loss": 0.016,
+      "step": 2821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917642385847523,
+      "loss": 0.014,
+      "step": 2822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917579023415641,
+      "loss": 0.0702,
+      "step": 2823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029175156373075135,
+      "loss": 0.0261,
+      "step": 2824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029174522275241996,
+      "loss": 0.0199,
+      "step": 2825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029173887940667583,
+      "loss": 0.0124,
+      "step": 2826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917325336936248,
+      "loss": 0.0096,
+      "step": 2827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917261856133729,
+      "loss": 0.0269,
+      "step": 2828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917198351660262,
+      "loss": 0.0068,
+      "step": 2829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029171348235169074,
+      "loss": 0.0097,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917071271704726,
+      "loss": 0.0127,
+      "step": 2831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000291700769622478,
+      "loss": 0.027,
+      "step": 2832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000291694409707813,
+      "loss": 0.0105,
+      "step": 2833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029168804742658394,
+      "loss": 0.0457,
+      "step": 2834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029168168277889697,
+      "loss": 0.0162,
+      "step": 2835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916753157648585,
+      "loss": 0.0957,
+      "step": 2836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029166894638457485,
+      "loss": 0.0051,
+      "step": 2837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029166257463815237,
+      "loss": 0.1087,
+      "step": 2838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029165620052569745,
+      "loss": 0.026,
+      "step": 2839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916498240473166,
+      "loss": 0.0133,
+      "step": 2840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916434452031163,
+      "loss": 0.0391,
+      "step": 2841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916370639932031,
+      "loss": 0.0102,
+      "step": 2842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916306804176836,
+      "loss": 0.0175,
+      "step": 2843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916242944766644,
+      "loss": 0.0191,
+      "step": 2844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029161790617025205,
+      "loss": 0.0119,
+      "step": 2845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916115154985534,
+      "loss": 0.01,
+      "step": 2846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029160512246167514,
+      "loss": 0.014,
+      "step": 2847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029159872705972407,
+      "loss": 0.0068,
+      "step": 2848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915923292928069,
+      "loss": 0.0419,
+      "step": 2849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029158592916103064,
+      "loss": 0.0491,
+      "step": 2850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029157952666450203,
+      "loss": 0.0129,
+      "step": 2851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915731218033281,
+      "loss": 0.0147,
+      "step": 2852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029156671457761577,
+      "loss": 0.0174,
+      "step": 2853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029156030498747213,
+      "loss": 0.0347,
+      "step": 2854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029155389303300416,
+      "loss": 0.0397,
+      "step": 2855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029154747871431896,
+      "loss": 0.0497,
+      "step": 2856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915410620315237,
+      "loss": 0.003,
+      "step": 2857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915346429847255,
+      "loss": 0.0093,
+      "step": 2858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029152822157403163,
+      "loss": 0.0431,
+      "step": 2859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915217977995493,
+      "loss": 0.0113,
+      "step": 2860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029151537166138575,
+      "loss": 0.0113,
+      "step": 2861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915089431596484,
+      "loss": 0.0899,
+      "step": 2862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029150251229444457,
+      "loss": 0.0156,
+      "step": 2863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029149607906588175,
+      "loss": 0.0222,
+      "step": 2864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029148964347406724,
+      "loss": 0.0086,
+      "step": 2865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029148320551910866,
+      "loss": 0.021,
+      "step": 2866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029147676520111345,
+      "loss": 0.0175,
+      "step": 2867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914703225201892,
+      "loss": 0.0152,
+      "step": 2868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914638774764435,
+      "loss": 0.0286,
+      "step": 2869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029145743006998403,
+      "loss": 0.0165,
+      "step": 2870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029145098030091847,
+      "loss": 0.0192,
+      "step": 2871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914445281693545,
+      "loss": 0.0265,
+      "step": 2872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029143807367539997,
+      "loss": 0.0256,
+      "step": 2873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002914316168191626,
+      "loss": 0.0402,
+      "step": 2874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029142515760075026,
+      "loss": 0.0456,
+      "step": 2875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002914186960202708,
+      "loss": 0.0279,
+      "step": 2876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002914122320778322,
+      "loss": 0.0203,
+      "step": 2877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002914057657735424,
+      "loss": 0.0468,
+      "step": 2878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029139929710750935,
+      "loss": 0.0429,
+      "step": 2879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029139282607984116,
+      "loss": 0.0237,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029138635269064587,
+      "loss": 0.0151,
+      "step": 2881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029137987694003163,
+      "loss": 0.0472,
+      "step": 2882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029137339882810654,
+      "loss": 0.0388,
+      "step": 2883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913669183549788,
+      "loss": 0.0182,
+      "step": 2884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913604355207567,
+      "loss": 0.0134,
+      "step": 2885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029135395032554856,
+      "loss": 0.007,
+      "step": 2886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913474627694625,
+      "loss": 0.0274,
+      "step": 2887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913409728526071,
+      "loss": 0.0238,
+      "step": 2888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913344805750906,
+      "loss": 0.0194,
+      "step": 2889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029132798593702145,
+      "loss": 0.0155,
+      "step": 2890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029132148893850826,
+      "loss": 0.0394,
+      "step": 2891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029131498957965936,
+      "loss": 0.0208,
+      "step": 2892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029130848786058347,
+      "loss": 0.0248,
+      "step": 2893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029130198378138903,
+      "loss": 0.009,
+      "step": 2894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912954773421847,
+      "loss": 0.0138,
+      "step": 2895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029128896854307924,
+      "loss": 0.0316,
+      "step": 2896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029128245738418133,
+      "loss": 0.0201,
+      "step": 2897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029127594386559963,
+      "loss": 0.0188,
+      "step": 2898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000291269427987443,
+      "loss": 0.0448,
+      "step": 2899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912629097498203,
+      "loss": 0.0563,
+      "step": 2900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912563891528403,
+      "loss": 0.0319,
+      "step": 2901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000291249866196612,
+      "loss": 0.0149,
+      "step": 2902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912433408812443,
+      "loss": 0.0439,
+      "step": 2903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029123681320684616,
+      "loss": 0.0026,
+      "step": 2904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912302831735267,
+      "loss": 0.0309,
+      "step": 2905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029122375078139486,
+      "loss": 0.027,
+      "step": 2906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029121721603055984,
+      "loss": 0.0173,
+      "step": 2907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912106789211307,
+      "loss": 0.0127,
+      "step": 2908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912041394532167,
+      "loss": 0.0042,
+      "step": 2909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029119759762692704,
+      "loss": 0.0128,
+      "step": 2910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029119105344237097,
+      "loss": 0.0189,
+      "step": 2911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029118450689965777,
+      "loss": 0.0306,
+      "step": 2912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911779579988968,
+      "loss": 0.0092,
+      "step": 2913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029117140674019747,
+      "loss": 0.0135,
+      "step": 2914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911648531236691,
+      "loss": 0.0108,
+      "step": 2915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029115829714942125,
+      "loss": 0.0181,
+      "step": 2916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911517388175634,
+      "loss": 0.0285,
+      "step": 2917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911451781282051,
+      "loss": 0.0316,
+      "step": 2918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029113861508145585,
+      "loss": 0.0279,
+      "step": 2919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911320496774253,
+      "loss": 0.0117,
+      "step": 2920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029112548191622317,
+      "loss": 0.01,
+      "step": 2921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000291118911797959,
+      "loss": 0.017,
+      "step": 2922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029111233932274276,
+      "loss": 0.007,
+      "step": 2923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029110576449068397,
+      "loss": 0.0657,
+      "step": 2924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910991873018926,
+      "loss": 0.0554,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029109260775647846,
+      "loss": 0.0371,
+      "step": 2926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029108602585455143,
+      "loss": 0.0396,
+      "step": 2927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910794415962214,
+      "loss": 0.0055,
+      "step": 2928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910728549815985,
+      "loss": 0.0238,
+      "step": 2929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029106626601079253,
+      "loss": 0.0162,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910596746839137,
+      "loss": 0.0085,
+      "step": 2931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000291053081001072,
+      "loss": 0.0094,
+      "step": 2932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910464849623776,
+      "loss": 0.0296,
+      "step": 2933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029103988656794063,
+      "loss": 0.0431,
+      "step": 2934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029103328581787133,
+      "loss": 0.0188,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029102668271228,
+      "loss": 0.015,
+      "step": 2936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910200772512768,
+      "loss": 0.0189,
+      "step": 2937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029101346943497214,
+      "loss": 0.0067,
+      "step": 2938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029100685926347636,
+      "loss": 0.005,
+      "step": 2939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029100024673689987,
+      "loss": 0.0291,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909936318553531,
+      "loss": 0.0186,
+      "step": 2941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029098701461894645,
+      "loss": 0.0077,
+      "step": 2942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909803950277906,
+      "loss": 0.0325,
+      "step": 2943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909737730819961,
+      "loss": 0.0479,
+      "step": 2944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909671487816734,
+      "loss": 0.0362,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909605221269332,
+      "loss": 0.0425,
+      "step": 2946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909538931178862,
+      "loss": 0.0163,
+      "step": 2947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909472617546432,
+      "loss": 0.0154,
+      "step": 2948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909406280373148,
+      "loss": 0.0151,
+      "step": 2949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909339919660119,
+      "loss": 0.0571,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029092735354084527,
+      "loss": 0.0428,
+      "step": 2951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909207127619259,
+      "loss": 0.0141,
+      "step": 2952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909140696293645,
+      "loss": 0.024,
+      "step": 2953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909074241432722,
+      "loss": 0.0259,
+      "step": 2954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909007763037599,
+      "loss": 0.0159,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908941261109387,
+      "loss": 0.0348,
+      "step": 2956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029088747356491963,
+      "loss": 0.0213,
+      "step": 2957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029088081866581383,
+      "loss": 0.0143,
+      "step": 2958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908741614137324,
+      "loss": 0.0315,
+      "step": 2959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908675018087866,
+      "loss": 0.0597,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908608398510876,
+      "loss": 0.0103,
+      "step": 2961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908541755407466,
+      "loss": 0.0255,
+      "step": 2962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029084750887787506,
+      "loss": 0.0228,
+      "step": 2963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029084083986258426,
+      "loss": 0.0246,
+      "step": 2964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029083416849498556,
+      "loss": 0.0129,
+      "step": 2965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908274947751904,
+      "loss": 0.0165,
+      "step": 2966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908208187033103,
+      "loss": 0.0096,
+      "step": 2967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908141402794566,
+      "loss": 0.0124,
+      "step": 2968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000290807459503741,
+      "loss": 0.0327,
+      "step": 2969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029080077637627503,
+      "loss": 0.0241,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029079409089717036,
+      "loss": 0.0234,
+      "step": 2971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029078740306653857,
+      "loss": 0.015,
+      "step": 2972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907807128844914,
+      "loss": 0.0166,
+      "step": 2973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907740203511406,
+      "loss": 0.0084,
+      "step": 2974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907673254665979,
+      "loss": 0.0213,
+      "step": 2975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029076062823097515,
+      "loss": 0.0575,
+      "step": 2976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907539286443842,
+      "loss": 0.0092,
+      "step": 2977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029074722670693697,
+      "loss": 0.0483,
+      "step": 2978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029074052241874536,
+      "loss": 0.0283,
+      "step": 2979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029073381577992137,
+      "loss": 0.0211,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000290727106790577,
+      "loss": 0.0238,
+      "step": 2981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907203954508243,
+      "loss": 0.0277,
+      "step": 2982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907136817607754,
+      "loss": 0.0094,
+      "step": 2983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029070696572054233,
+      "loss": 0.0161,
+      "step": 2984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029070024733023736,
+      "loss": 0.0129,
+      "step": 2985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906935265899727,
+      "loss": 0.0189,
+      "step": 2986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906868034998606,
+      "loss": 0.0132,
+      "step": 2987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906800780600132,
+      "loss": 0.0172,
+      "step": 2988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000290673350270543,
+      "loss": 0.0281,
+      "step": 2989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029066662013156234,
+      "loss": 0.0247,
+      "step": 2990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029065988764318353,
+      "loss": 0.009,
+      "step": 2991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029065315280551914,
+      "loss": 0.0121,
+      "step": 2992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906464156186816,
+      "loss": 0.0118,
+      "step": 2993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906396760827834,
+      "loss": 0.0375,
+      "step": 2994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029063293419793714,
+      "loss": 0.0147,
+      "step": 2995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906261899642554,
+      "loss": 0.0331,
+      "step": 2996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029061944338185086,
+      "loss": 0.0143,
+      "step": 2997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906126944508362,
+      "loss": 0.0071,
+      "step": 2998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029060594317132407,
+      "loss": 0.0135,
+      "step": 2999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905991895434273,
+      "loss": 0.0219,
+      "step": 3000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029059243356725865,
+      "loss": 0.0141,
+      "step": 3001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000290585675242931,
+      "loss": 0.0215,
+      "step": 3002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905789145705572,
+      "loss": 0.0212,
+      "step": 3003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029057215155025017,
+      "loss": 0.0483,
+      "step": 3004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905653861821228,
+      "loss": 0.0504,
+      "step": 3005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029055861846628817,
+      "loss": 0.1316,
+      "step": 3006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905518484028593,
+      "loss": 0.0184,
+      "step": 3007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905450759919493,
+      "loss": 0.0347,
+      "step": 3008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029053830123367116,
+      "loss": 0.0112,
+      "step": 3009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029053152412813815,
+      "loss": 0.0318,
+      "step": 3010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905247446754634,
+      "loss": 0.0673,
+      "step": 3011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029051796287576015,
+      "loss": 0.0099,
+      "step": 3012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905111787291417,
+      "loss": 0.0365,
+      "step": 3013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905043922357213,
+      "loss": 0.0116,
+      "step": 3014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002904976033956123,
+      "loss": 0.0509,
+      "step": 3015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029049081220892817,
+      "loss": 0.0159,
+      "step": 3016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029048401867578225,
+      "loss": 0.0346,
+      "step": 3017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029047722279628807,
+      "loss": 0.0216,
+      "step": 3018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002904704245705591,
+      "loss": 0.0192,
+      "step": 3019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002904636239987088,
+      "loss": 0.015,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904568210808509,
+      "loss": 0.0135,
+      "step": 3021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029045001581709897,
+      "loss": 0.0152,
+      "step": 3022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904432082075666,
+      "loss": 0.0149,
+      "step": 3023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029043639825236764,
+      "loss": 0.0732,
+      "step": 3024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029042958595161566,
+      "loss": 0.0508,
+      "step": 3025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029042277130542456,
+      "loss": 0.007,
+      "step": 3026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904159543139081,
+      "loss": 0.0442,
+      "step": 3027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029040913497718014,
+      "loss": 0.0252,
+      "step": 3028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029040231329535455,
+      "loss": 0.0413,
+      "step": 3029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029039548926854537,
+      "loss": 0.0268,
+      "step": 3030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903886628968665,
+      "loss": 0.0255,
+      "step": 3031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903818341804319,
+      "loss": 0.0329,
+      "step": 3032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903750031193557,
+      "loss": 0.0059,
+      "step": 3033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000290368169713752,
+      "loss": 0.0361,
+      "step": 3034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029036133396373486,
+      "loss": 0.0128,
+      "step": 3035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029035449586941854,
+      "loss": 0.0292,
+      "step": 3036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029034765543091716,
+      "loss": 0.0099,
+      "step": 3037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000290340812648345,
+      "loss": 0.0369,
+      "step": 3038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903339675218164,
+      "loss": 0.0276,
+      "step": 3039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903271200514456,
+      "loss": 0.0171,
+      "step": 3040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029032027023734704,
+      "loss": 0.017,
+      "step": 3041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903134180796351,
+      "loss": 0.0058,
+      "step": 3042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903065635784242,
+      "loss": 0.0349,
+      "step": 3043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029029970673382884,
+      "loss": 0.0074,
+      "step": 3044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902928475459635,
+      "loss": 0.0143,
+      "step": 3045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029028598601494286,
+      "loss": 0.0402,
+      "step": 3046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029027912214088145,
+      "loss": 0.015,
+      "step": 3047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902722559238939,
+      "loss": 0.0229,
+      "step": 3048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902653873640948,
+      "loss": 0.016,
+      "step": 3049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029025851646159907,
+      "loss": 0.0396,
+      "step": 3050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902516432165213,
+      "loss": 0.0177,
+      "step": 3051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029024476762897633,
+      "loss": 0.0377,
+      "step": 3052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029023788969907905,
+      "loss": 0.0042,
+      "step": 3053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029023100942694426,
+      "loss": 0.0443,
+      "step": 3054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000290224126812687,
+      "loss": 0.0166,
+      "step": 3055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029021724185642205,
+      "loss": 0.0256,
+      "step": 3056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902103545582645,
+      "loss": 0.0253,
+      "step": 3057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902034649183293,
+      "loss": 0.0131,
+      "step": 3058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029019657293673164,
+      "loss": 0.0416,
+      "step": 3059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901896786135866,
+      "loss": 0.0162,
+      "step": 3060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901827819490093,
+      "loss": 0.0115,
+      "step": 3061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029017588294311483,
+      "loss": 0.0715,
+      "step": 3062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029016898159601865,
+      "loss": 0.0698,
+      "step": 3063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901620779078358,
+      "loss": 0.0402,
+      "step": 3064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901551718786817,
+      "loss": 0.0321,
+      "step": 3065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029014826350867164,
+      "loss": 0.0178,
+      "step": 3066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901413527979211,
+      "loss": 0.0297,
+      "step": 3067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901344397465454,
+      "loss": 0.0267,
+      "step": 3068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029012752435466,
+      "loss": 0.0173,
+      "step": 3069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029012060662238047,
+      "loss": 0.0152,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002901136865498223,
+      "loss": 0.0532,
+      "step": 3071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000290106764137101,
+      "loss": 0.0113,
+      "step": 3072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029009983938433243,
+      "loss": 0.0264,
+      "step": 3073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000290092912291632,
+      "loss": 0.0678,
+      "step": 3074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900859828591155,
+      "loss": 0.0223,
+      "step": 3075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029007905108689865,
+      "loss": 0.0273,
+      "step": 3076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029007211697509725,
+      "loss": 0.0485,
+      "step": 3077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000290065180523827,
+      "loss": 0.047,
+      "step": 3078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900582417332039,
+      "loss": 0.0339,
+      "step": 3079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900513006033438,
+      "loss": 0.0207,
+      "step": 3080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029004435713436265,
+      "loss": 0.0299,
+      "step": 3081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900374113263763,
+      "loss": 0.0162,
+      "step": 3082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000290030463179501,
+      "loss": 0.0287,
+      "step": 3083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029002351269385247,
+      "loss": 0.0451,
+      "step": 3084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029001655986954703,
+      "loss": 0.0629,
+      "step": 3085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900096047067007,
+      "loss": 0.0286,
+      "step": 3086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029000264720542974,
+      "loss": 0.0537,
+      "step": 3087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899956873658503,
+      "loss": 0.0044,
+      "step": 3088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028998872518807856,
+      "loss": 0.0574,
+      "step": 3089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899817606722309,
+      "loss": 0.0082,
+      "step": 3090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899747938184236,
+      "loss": 0.0194,
+      "step": 3091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899678246267731,
+      "loss": 0.0197,
+      "step": 3092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899608530973956,
+      "loss": 0.0095,
+      "step": 3093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899538792304077,
+      "loss": 0.0134,
+      "step": 3094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899469030259258,
+      "loss": 0.0114,
+      "step": 3095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899399244840665,
+      "loss": 0.0372,
+      "step": 3096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028993294360494626,
+      "loss": 0.0084,
+      "step": 3097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899259603886818,
+      "loss": 0.0954,
+      "step": 3098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899189748353896,
+      "loss": 0.01,
+      "step": 3099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028991198694518645,
+      "loss": 0.01,
+      "step": 3100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028990499671818895,
+      "loss": 0.0165,
+      "step": 3101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289898004154514,
+      "loss": 0.0311,
+      "step": 3102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028989100925427825,
+      "loss": 0.0078,
+      "step": 3103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028988401201759863,
+      "loss": 0.018,
+      "step": 3104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028987701244459187,
+      "loss": 0.0192,
+      "step": 3105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028987001053537505,
+      "loss": 0.0311,
+      "step": 3106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289863006290065,
+      "loss": 0.011,
+      "step": 3107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028985599970877875,
+      "loss": 0.0264,
+      "step": 3108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898489907916333,
+      "loss": 0.0184,
+      "step": 3109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028984197953874573,
+      "loss": 0.0113,
+      "step": 3110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028983496595023315,
+      "loss": 0.0244,
+      "step": 3111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028982795002621263,
+      "loss": 0.0568,
+      "step": 3112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898209317668014,
+      "loss": 0.0224,
+      "step": 3113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898139111721167,
+      "loss": 0.0727,
+      "step": 3114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898068882422757,
+      "loss": 0.0464,
+      "step": 3115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002897998629773959,
+      "loss": 0.0111,
+      "step": 3116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002897928353775944,
+      "loss": 0.0196,
+      "step": 3117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002897858054429887,
+      "loss": 0.0126,
+      "step": 3118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028977877317369616,
+      "loss": 0.0146,
+      "step": 3119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897717385698343,
+      "loss": 0.0364,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028976470163152054,
+      "loss": 0.0364,
+      "step": 3121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897576623588724,
+      "loss": 0.0241,
+      "step": 3122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028975062075200754,
+      "loss": 0.0342,
+      "step": 3123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897435768110435,
+      "loss": 0.0253,
+      "step": 3124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289736530536098,
+      "loss": 0.0117,
+      "step": 3125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897294819272886,
+      "loss": 0.0093,
+      "step": 3126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897224309847331,
+      "loss": 0.0373,
+      "step": 3127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897153777085493,
+      "loss": 0.0122,
+      "step": 3128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897083220988549,
+      "loss": 0.0232,
+      "step": 3129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028970126415576785,
+      "loss": 0.0447,
+      "step": 3130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289694203879406,
+      "loss": 0.0219,
+      "step": 3131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896871412698872,
+      "loss": 0.0104,
+      "step": 3132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028968007632732944,
+      "loss": 0.0567,
+      "step": 3133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896730090518508,
+      "loss": 0.0317,
+      "step": 3134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028966593944356925,
+      "loss": 0.0153,
+      "step": 3135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028965886750260283,
+      "loss": 0.0517,
+      "step": 3136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028965179322906976,
+      "loss": 0.034,
+      "step": 3137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896447166230881,
+      "loss": 0.0411,
+      "step": 3138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028963763768477606,
+      "loss": 0.0072,
+      "step": 3139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028963055641425194,
+      "loss": 0.0287,
+      "step": 3140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896234728116339,
+      "loss": 0.0126,
+      "step": 3141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896163868770403,
+      "loss": 0.0057,
+      "step": 3142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028960929861058957,
+      "loss": 0.042,
+      "step": 3143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028960220801239997,
+      "loss": 0.0142,
+      "step": 3144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028959511508258996,
+      "loss": 0.0125,
+      "step": 3145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028958801982127805,
+      "loss": 0.042,
+      "step": 3146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895809222285827,
+      "loss": 0.034,
+      "step": 3147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028957382230462245,
+      "loss": 0.0203,
+      "step": 3148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289566720049516,
+      "loss": 0.015,
+      "step": 3149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028955961546338175,
+      "loss": 0.0106,
+      "step": 3150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028955250854633855,
+      "loss": 0.0119,
+      "step": 3151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028954539929850506,
+      "loss": 0.0107,
+      "step": 3152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028953828772,
+      "loss": 0.0357,
+      "step": 3153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028953117381094203,
+      "loss": 0.0199,
+      "step": 3154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028952405757145015,
+      "loss": 0.0257,
+      "step": 3155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895169390016431,
+      "loss": 0.0315,
+      "step": 3156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895098181016399,
+      "loss": 0.0195,
+      "step": 3157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028950269487155935,
+      "loss": 0.0496,
+      "step": 3158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894955693115204,
+      "loss": 0.036,
+      "step": 3159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894884414216422,
+      "loss": 0.0244,
+      "step": 3160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894813112020437,
+      "loss": 0.0186,
+      "step": 3161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289474178652844,
+      "loss": 0.0096,
+      "step": 3162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028946704377416226,
+      "loss": 0.0199,
+      "step": 3163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028945990656611763,
+      "loss": 0.0544,
+      "step": 3164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028945276702882933,
+      "loss": 0.0235,
+      "step": 3165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028944562516241656,
+      "loss": 0.0126,
+      "step": 3166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028943848096699864,
+      "loss": 0.0433,
+      "step": 3167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894313344426948,
+      "loss": 0.0126,
+      "step": 3168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028942418558962455,
+      "loss": 0.017,
+      "step": 3169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894170344079072,
+      "loss": 0.0247,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894098808976623,
+      "loss": 0.0365,
+      "step": 3171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894027250590091,
+      "loss": 0.0306,
+      "step": 3172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893955668920673,
+      "loss": 0.0558,
+      "step": 3173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893884063969564,
+      "loss": 0.0386,
+      "step": 3174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289381243573796,
+      "loss": 0.0162,
+      "step": 3175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893740784227058,
+      "loss": 0.015,
+      "step": 3176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893669109438053,
+      "loss": 0.0206,
+      "step": 3177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028935974113721436,
+      "loss": 0.034,
+      "step": 3178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028935256900305273,
+      "loss": 0.007,
+      "step": 3179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893453945414401,
+      "loss": 0.0231,
+      "step": 3180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893382177524963,
+      "loss": 0.0144,
+      "step": 3181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893310386363413,
+      "loss": 0.0316,
+      "step": 3182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028932385719309496,
+      "loss": 0.0454,
+      "step": 3183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028931667342287716,
+      "loss": 0.0188,
+      "step": 3184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289309487325808,
+      "loss": 0.0036,
+      "step": 3185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893022989020074,
+      "loss": 0.036,
+      "step": 3186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028929510815159544,
+      "loss": 0.0337,
+      "step": 3187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892879150746923,
+      "loss": 0.0112,
+      "step": 3188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289280719671418,
+      "loss": 0.0275,
+      "step": 3189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892735219418928,
+      "loss": 0.029,
+      "step": 3190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028926632188623684,
+      "loss": 0.0075,
+      "step": 3191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028925911950457045,
+      "loss": 0.025,
+      "step": 3192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892519147970139,
+      "loss": 0.0435,
+      "step": 3193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892447077636875,
+      "loss": 0.027,
+      "step": 3194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892374984047116,
+      "loss": 0.0072,
+      "step": 3195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028923028672020673,
+      "loss": 0.034,
+      "step": 3196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892230727102932,
+      "loss": 0.0175,
+      "step": 3197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892158563750916,
+      "loss": 0.0152,
+      "step": 3198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892086377147224,
+      "loss": 0.0144,
+      "step": 3199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028920141672930617,
+      "loss": 0.0476,
+      "step": 3200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891941934189635,
+      "loss": 0.0161,
+      "step": 3201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028918696778381503,
+      "loss": 0.0315,
+      "step": 3202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891797398239815,
+      "loss": 0.0385,
+      "step": 3203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891725095395836,
+      "loss": 0.011,
+      "step": 3204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891652769307421,
+      "loss": 0.0158,
+      "step": 3205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028915804199757775,
+      "loss": 0.0195,
+      "step": 3206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891508047402114,
+      "loss": 0.0207,
+      "step": 3207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289143565158764,
+      "loss": 0.0525,
+      "step": 3208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891363232533564,
+      "loss": 0.0121,
+      "step": 3209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028912907902410947,
+      "loss": 0.0389,
+      "step": 3210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891218324711444,
+      "loss": 0.0159,
+      "step": 3211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028911458359458204,
+      "loss": 0.0228,
+      "step": 3212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891073323945436,
+      "loss": 0.0496,
+      "step": 3213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891000788711501,
+      "loss": 0.0294,
+      "step": 3214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028909282302452274,
+      "loss": 0.0091,
+      "step": 3215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028908556485478265,
+      "loss": 0.0462,
+      "step": 3216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028907830436205115,
+      "loss": 0.0139,
+      "step": 3217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028907104154644934,
+      "loss": 0.0271,
+      "step": 3218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028906377640809867,
+      "loss": 0.0237,
+      "step": 3219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890565089471204,
+      "loss": 0.0382,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000289049239163636,
+      "loss": 0.0225,
+      "step": 3221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890419670577668,
+      "loss": 0.0255,
+      "step": 3222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890346926296343,
+      "loss": 0.025,
+      "step": 3223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028902741587935994,
+      "loss": 0.043,
+      "step": 3224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890201368070653,
+      "loss": 0.0148,
+      "step": 3225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028901285541287205,
+      "loss": 0.0475,
+      "step": 3226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890055716969016,
+      "loss": 0.0174,
+      "step": 3227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889982856592758,
+      "loss": 0.0088,
+      "step": 3228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889909973001162,
+      "loss": 0.0548,
+      "step": 3229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889837066195446,
+      "loss": 0.0299,
+      "step": 3230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889764136176827,
+      "loss": 0.0044,
+      "step": 3231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889691182946525,
+      "loss": 0.0326,
+      "step": 3232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028896182065057556,
+      "loss": 0.0088,
+      "step": 3233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028895452068557393,
+      "loss": 0.0066,
+      "step": 3234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889472183997696,
+      "loss": 0.0298,
+      "step": 3235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889399137932843,
+      "loss": 0.0069,
+      "step": 3236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028893260686624025,
+      "loss": 0.0525,
+      "step": 3237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889252976187594,
+      "loss": 0.0324,
+      "step": 3238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028891798605096383,
+      "loss": 0.0306,
+      "step": 3239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889106721629757,
+      "loss": 0.0465,
+      "step": 3240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028890335595491715,
+      "loss": 0.0207,
+      "step": 3241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888960374269103,
+      "loss": 0.009,
+      "step": 3242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028888871657907744,
+      "loss": 0.0399,
+      "step": 3243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028888139341154087,
+      "loss": 0.0386,
+      "step": 3244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028887406792442287,
+      "loss": 0.0082,
+      "step": 3245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888667401178458,
+      "loss": 0.0191,
+      "step": 3246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000288859409991932,
+      "loss": 0.0256,
+      "step": 3247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028885207754680397,
+      "loss": 0.0436,
+      "step": 3248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028884474278258415,
+      "loss": 0.0431,
+      "step": 3249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000288837405699395,
+      "loss": 0.0357,
+      "step": 3250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028883006629735916,
+      "loss": 0.0113,
+      "step": 3251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888227245765991,
+      "loss": 0.0112,
+      "step": 3252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888153805372375,
+      "loss": 0.0095,
+      "step": 3253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028880803417939704,
+      "loss": 0.0104,
+      "step": 3254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888006855032004,
+      "loss": 0.0137,
+      "step": 3255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028879333450877026,
+      "loss": 0.0389,
+      "step": 3256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887859811962294,
+      "loss": 0.0298,
+      "step": 3257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887786255657008,
+      "loss": 0.0242,
+      "step": 3258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887712676173071,
+      "loss": 0.0072,
+      "step": 3259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028876390735117134,
+      "loss": 0.0137,
+      "step": 3260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887565447674164,
+      "loss": 0.0258,
+      "step": 3261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887491798661652,
+      "loss": 0.0241,
+      "step": 3262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028874181264754075,
+      "loss": 0.0165,
+      "step": 3263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887344431116662,
+      "loss": 0.0435,
+      "step": 3264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028872707125866457,
+      "loss": 0.0186,
+      "step": 3265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028871969708865893,
+      "loss": 0.0461,
+      "step": 3266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002887123206017725,
+      "loss": 0.0331,
+      "step": 3267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002887049417981285,
+      "loss": 0.0171,
+      "step": 3268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886975606778502,
+      "loss": 0.019,
+      "step": 3269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886901772410608,
+      "loss": 0.0225,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886827914878836,
+      "loss": 0.0143,
+      "step": 3271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028867540341844205,
+      "loss": 0.0211,
+      "step": 3272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028866801303285947,
+      "loss": 0.0299,
+      "step": 3273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028866062033125933,
+      "loss": 0.0116,
+      "step": 3274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028865322531376503,
+      "loss": 0.0243,
+      "step": 3275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886458279805002,
+      "loss": 0.0292,
+      "step": 3276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886384283315883,
+      "loss": 0.037,
+      "step": 3277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288631026367153,
+      "loss": 0.0231,
+      "step": 3278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028862362208731786,
+      "loss": 0.0157,
+      "step": 3279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886162154922065,
+      "loss": 0.0099,
+      "step": 3280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028860880658194277,
+      "loss": 0.0284,
+      "step": 3281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028860139535665023,
+      "loss": 0.027,
+      "step": 3282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028859398181645287,
+      "loss": 0.0381,
+      "step": 3283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885865659614743,
+      "loss": 0.0151,
+      "step": 3284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885791477918385,
+      "loss": 0.0661,
+      "step": 3285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885717273076694,
+      "loss": 0.0441,
+      "step": 3286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028856430450909085,
+      "loss": 0.0101,
+      "step": 3287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885568793962268,
+      "loss": 0.0172,
+      "step": 3288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885494519692014,
+      "loss": 0.0318,
+      "step": 3289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028854202222813863,
+      "loss": 0.018,
+      "step": 3290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885345901731625,
+      "loss": 0.0338,
+      "step": 3291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885271558043973,
+      "loss": 0.0721,
+      "step": 3292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288519719121967,
+      "loss": 0.0119,
+      "step": 3293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288512280125996,
+      "loss": 0.045,
+      "step": 3294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028850483881660843,
+      "loss": 0.0184,
+      "step": 3295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884973951939286,
+      "loss": 0.0192,
+      "step": 3296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884899492580809,
+      "loss": 0.0291,
+      "step": 3297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028848250100918955,
+      "loss": 0.0417,
+      "step": 3298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288475050447379,
+      "loss": 0.0296,
+      "step": 3299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884675975727738,
+      "loss": 0.0263,
+      "step": 3300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028846014238549834,
+      "loss": 0.0124,
+      "step": 3301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884526848856771,
+      "loss": 0.0101,
+      "step": 3302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884452250734347,
+      "loss": 0.045,
+      "step": 3303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884377629488957,
+      "loss": 0.0144,
+      "step": 3304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028843029851218475,
+      "loss": 0.0238,
+      "step": 3305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884228317634265,
+      "loss": 0.048,
+      "step": 3306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028841536270274564,
+      "loss": 0.037,
+      "step": 3307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288407891330267,
+      "loss": 0.0333,
+      "step": 3308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884004176461153,
+      "loss": 0.008,
+      "step": 3309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028839294165041536,
+      "loss": 0.0157,
+      "step": 3310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288385463343292,
+      "loss": 0.0455,
+      "step": 3311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028837798272487026,
+      "loss": 0.0224,
+      "step": 3312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288370499795275,
+      "loss": 0.0386,
+      "step": 3313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002883630145546312,
+      "loss": 0.0497,
+      "step": 3314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028835552700306387,
+      "loss": 0.0189,
+      "step": 3315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288348037140698,
+      "loss": 0.0287,
+      "step": 3316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883405449676589,
+      "loss": 0.0126,
+      "step": 3317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028833305048407145,
+      "loss": 0.0318,
+      "step": 3318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028832555369006095,
+      "loss": 0.0134,
+      "step": 3319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883180545857526,
+      "loss": 0.0528,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883105531712717,
+      "loss": 0.0276,
+      "step": 3321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883030494467435,
+      "loss": 0.0226,
+      "step": 3322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882955434122932,
+      "loss": 0.0125,
+      "step": 3323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028828803506804633,
+      "loss": 0.0212,
+      "step": 3324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882805244141282,
+      "loss": 0.0126,
+      "step": 3325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028827301145066433,
+      "loss": 0.0256,
+      "step": 3326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882654961777802,
+      "loss": 0.0282,
+      "step": 3327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882579785956012,
+      "loss": 0.0063,
+      "step": 3328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288250458704253,
+      "loss": 0.0377,
+      "step": 3329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882429365038612,
+      "loss": 0.0112,
+      "step": 3330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882354119945514,
+      "loss": 0.0236,
+      "step": 3331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882278851764493,
+      "loss": 0.0298,
+      "step": 3332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028822035604968057,
+      "loss": 0.0493,
+      "step": 3333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288212824614371,
+      "loss": 0.029,
+      "step": 3334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028820529087064634,
+      "loss": 0.0333,
+      "step": 3335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881977548186325,
+      "loss": 0.0217,
+      "step": 3336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881902164584552,
+      "loss": 0.0378,
+      "step": 3337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028818267579024044,
+      "loss": 0.0131,
+      "step": 3338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881751328141142,
+      "loss": 0.0064,
+      "step": 3339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881675875302024,
+      "loss": 0.0075,
+      "step": 3340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028816003993863116,
+      "loss": 0.0314,
+      "step": 3341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028815249003952633,
+      "loss": 0.0084,
+      "step": 3342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881449378330142,
+      "loss": 0.016,
+      "step": 3343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881373833192208,
+      "loss": 0.0385,
+      "step": 3344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881298264982724,
+      "loss": 0.0116,
+      "step": 3345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028812226737029515,
+      "loss": 0.0148,
+      "step": 3346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881147059354153,
+      "loss": 0.0259,
+      "step": 3347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028810714219375917,
+      "loss": 0.0233,
+      "step": 3348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288099576145453,
+      "loss": 0.0068,
+      "step": 3349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028809200779062334,
+      "loss": 0.0226,
+      "step": 3350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880844371293964,
+      "loss": 0.0444,
+      "step": 3351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880768641618987,
+      "loss": 0.0265,
+      "step": 3352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028806928888825675,
+      "loss": 0.0436,
+      "step": 3353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028806171130859703,
+      "loss": 0.0088,
+      "step": 3354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880541314230462,
+      "loss": 0.0166,
+      "step": 3355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880465492317307,
+      "loss": 0.0151,
+      "step": 3356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028803896473477726,
+      "loss": 0.0188,
+      "step": 3357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028803137793231257,
+      "loss": 0.0121,
+      "step": 3358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880237888244633,
+      "loss": 0.0367,
+      "step": 3359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028801619741135624,
+      "loss": 0.0167,
+      "step": 3360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880086036931181,
+      "loss": 0.0192,
+      "step": 3361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880010076698758,
+      "loss": 0.0136,
+      "step": 3362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002879934093417562,
+      "loss": 0.0205,
+      "step": 3363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002879858087088862,
+      "loss": 0.0184,
+      "step": 3364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879782057713927,
+      "loss": 0.0027,
+      "step": 3365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879706005294027,
+      "loss": 0.0185,
+      "step": 3366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028796299298304324,
+      "loss": 0.0763,
+      "step": 3367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028795538313244136,
+      "loss": 0.0107,
+      "step": 3368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879477709777242,
+      "loss": 0.0089,
+      "step": 3369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028794015651901887,
+      "loss": 0.0438,
+      "step": 3370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879325397564525,
+      "loss": 0.0259,
+      "step": 3371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879249206901524,
+      "loss": 0.0086,
+      "step": 3372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028791729932024573,
+      "loss": 0.0162,
+      "step": 3373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028790967564685984,
+      "loss": 0.0309,
+      "step": 3374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028790204967012206,
+      "loss": 0.0196,
+      "step": 3375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878944213901597,
+      "loss": 0.0658,
+      "step": 3376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878867908071003,
+      "loss": 0.0288,
+      "step": 3377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028787915792107113,
+      "loss": 0.0155,
+      "step": 3378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878715227321998,
+      "loss": 0.0074,
+      "step": 3379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028786388524061374,
+      "loss": 0.0048,
+      "step": 3380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878562454464406,
+      "loss": 0.0148,
+      "step": 3381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028784860334980794,
+      "loss": 0.047,
+      "step": 3382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028784095895084336,
+      "loss": 0.0145,
+      "step": 3383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028783331224967466,
+      "loss": 0.0322,
+      "step": 3384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878256632464294,
+      "loss": 0.0238,
+      "step": 3385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878180119412354,
+      "loss": 0.0366,
+      "step": 3386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878103583342205,
+      "loss": 0.0726,
+      "step": 3387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878027024255124,
+      "loss": 0.0489,
+      "step": 3388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877950442152391,
+      "loss": 0.0065,
+      "step": 3389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877873837035284,
+      "loss": 0.0244,
+      "step": 3390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877797208905083,
+      "loss": 0.0332,
+      "step": 3391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028777205577630685,
+      "loss": 0.048,
+      "step": 3392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000287764388361052,
+      "loss": 0.0485,
+      "step": 3393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028775671864487176,
+      "loss": 0.027,
+      "step": 3394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877490466278943,
+      "loss": 0.0067,
+      "step": 3395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028774137231024784,
+      "loss": 0.0304,
+      "step": 3396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028773369569206036,
+      "loss": 0.0167,
+      "step": 3397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877260167734602,
+      "loss": 0.0322,
+      "step": 3398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028771833555457557,
+      "loss": 0.0365,
+      "step": 3399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028771065203553477,
+      "loss": 0.0105,
+      "step": 3400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028770296621646613,
+      "loss": 0.0451,
+      "step": 3401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000287695278097498,
+      "loss": 0.0106,
+      "step": 3402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028768758767875886,
+      "loss": 0.0618,
+      "step": 3403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876798949603771,
+      "loss": 0.0254,
+      "step": 3404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876721999424812,
+      "loss": 0.0127,
+      "step": 3405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028766450262519963,
+      "loss": 0.0307,
+      "step": 3406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876568030086611,
+      "loss": 0.0225,
+      "step": 3407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000287649101092994,
+      "loss": 0.0117,
+      "step": 3408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876413968783272,
+      "loss": 0.0411,
+      "step": 3409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876336903647892,
+      "loss": 0.0187,
+      "step": 3410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876259815525087,
+      "loss": 0.0092,
+      "step": 3411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028761827044161464,
+      "loss": 0.0094,
+      "step": 3412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028761055703223565,
+      "loss": 0.0563,
+      "step": 3413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002876028413245006,
+      "loss": 0.0251,
+      "step": 3414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875951233185383,
+      "loss": 0.0101,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028758740301447776,
+      "loss": 0.0242,
+      "step": 3416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875796804124479,
+      "loss": 0.0246,
+      "step": 3417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028757195551257767,
+      "loss": 0.0108,
+      "step": 3418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028756422831499606,
+      "loss": 0.0235,
+      "step": 3419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875564988198322,
+      "loss": 0.0184,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875487670272151,
+      "loss": 0.0374,
+      "step": 3421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287541032937274,
+      "loss": 0.0298,
+      "step": 3422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287533296550138,
+      "loss": 0.0242,
+      "step": 3423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028752555786593634,
+      "loss": 0.0341,
+      "step": 3424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875178168847983,
+      "loss": 0.0117,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875100736068531,
+      "loss": 0.0525,
+      "step": 3426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875023280322301,
+      "loss": 0.0484,
+      "step": 3427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028749458016105863,
+      "loss": 0.0184,
+      "step": 3428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874868299934682,
+      "loss": 0.0402,
+      "step": 3429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028747907752958815,
+      "loss": 0.0243,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028747132276954804,
+      "loss": 0.0127,
+      "step": 3431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028746356571347727,
+      "loss": 0.0194,
+      "step": 3432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874558063615055,
+      "loss": 0.0284,
+      "step": 3433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874480447137623,
+      "loss": 0.0192,
+      "step": 3434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028744028077037735,
+      "loss": 0.0166,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874325145314802,
+      "loss": 0.0541,
+      "step": 3436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874247459972007,
+      "loss": 0.0302,
+      "step": 3437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874169751676685,
+      "loss": 0.0371,
+      "step": 3438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874092020430134,
+      "loss": 0.0184,
+      "step": 3439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874014266233653,
+      "loss": 0.0311,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028739364890885394,
+      "loss": 0.0384,
+      "step": 3441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873858688996094,
+      "loss": 0.0712,
+      "step": 3442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028737808659576146,
+      "loss": 0.0169,
+      "step": 3443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028737030199744014,
+      "loss": 0.0264,
+      "step": 3444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873625151047755,
+      "loss": 0.0497,
+      "step": 3445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028735472591789754,
+      "loss": 0.0118,
+      "step": 3446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028734693443693637,
+      "loss": 0.0455,
+      "step": 3447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028733914066202213,
+      "loss": 0.0213,
+      "step": 3448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287331344593285,
+      "loss": 0.0189,
+      "step": 3449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873235462308552,
+      "loss": 0.0122,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287315745574863,
+      "loss": 0.0064,
+      "step": 3451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873079426254386,
+      "loss": 0.023,
+      "step": 3452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873001373827124,
+      "loss": 0.0232,
+      "step": 3453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872923298468147,
+      "loss": 0.0334,
+      "step": 3454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028728452001787594,
+      "loss": 0.0318,
+      "step": 3455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028727670789602657,
+      "loss": 0.0321,
+      "step": 3456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287268893481397,
+      "loss": 0.0495,
+      "step": 3457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872610767741179,
+      "loss": 0.0131,
+      "step": 3458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028725325777431964,
+      "loss": 0.0117,
+      "step": 3459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872454364821329,
+      "loss": 0.0504,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872376128976883,
+      "loss": 0.0153,
+      "step": 3461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028722978702111653,
+      "loss": 0.0157,
+      "step": 3462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002872219588525483,
+      "loss": 0.0387,
+      "step": 3463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028721412839211426,
+      "loss": 0.024,
+      "step": 3464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028720629563994533,
+      "loss": 0.0519,
+      "step": 3465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871984605961722,
+      "loss": 0.0148,
+      "step": 3466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871906232609259,
+      "loss": 0.0201,
+      "step": 3467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028718278363433714,
+      "loss": 0.0035,
+      "step": 3468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028717494171653696,
+      "loss": 0.0122,
+      "step": 3469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871670975076563,
+      "loss": 0.0394,
+      "step": 3470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871592510078262,
+      "loss": 0.0121,
+      "step": 3471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871514022171778,
+      "loss": 0.0073,
+      "step": 3472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000287143551135842,
+      "loss": 0.0368,
+      "step": 3473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028713569776395,
+      "loss": 0.0228,
+      "step": 3474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028712784210163303,
+      "loss": 0.0268,
+      "step": 3475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871199841490222,
+      "loss": 0.06,
+      "step": 3476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871121239062489,
+      "loss": 0.0228,
+      "step": 3477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871042613734442,
+      "loss": 0.016,
+      "step": 3478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870963965507396,
+      "loss": 0.0589,
+      "step": 3479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028708852943826635,
+      "loss": 0.0064,
+      "step": 3480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028708066003615593,
+      "loss": 0.0474,
+      "step": 3481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028707278834453975,
+      "loss": 0.0071,
+      "step": 3482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870649143635492,
+      "loss": 0.024,
+      "step": 3483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028705703809331584,
+      "loss": 0.025,
+      "step": 3484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028704915953397126,
+      "loss": 0.0223,
+      "step": 3485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000287041278685647,
+      "loss": 0.0076,
+      "step": 3486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028703339554847474,
+      "loss": 0.0261,
+      "step": 3487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870255101225861,
+      "loss": 0.0181,
+      "step": 3488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028701762240811277,
+      "loss": 0.0997,
+      "step": 3489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870097324051865,
+      "loss": 0.0172,
+      "step": 3490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870018401139391,
+      "loss": 0.0437,
+      "step": 3491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028699394553450235,
+      "loss": 0.0112,
+      "step": 3492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028698604866700807,
+      "loss": 0.0527,
+      "step": 3493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028697814951158824,
+      "loss": 0.0286,
+      "step": 3494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028697024806837476,
+      "loss": 0.011,
+      "step": 3495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002869623443374996,
+      "loss": 0.0113,
+      "step": 3496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028695443831909474,
+      "loss": 0.0125,
+      "step": 3497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028694653001329215,
+      "loss": 0.0268,
+      "step": 3498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028693861942022413,
+      "loss": 0.0227,
+      "step": 3499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028693070654002263,
+      "loss": 0.0228,
+      "step": 3500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002869227913728198,
+      "loss": 0.0535,
+      "step": 3501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000286914873918748,
+      "loss": 0.0084,
+      "step": 3502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002869069541779393,
+      "loss": 0.0097,
+      "step": 3503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002868990321505261,
+      "loss": 0.0087,
+      "step": 3504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028689110783664055,
+      "loss": 0.0126,
+      "step": 3505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028688318123641513,
+      "loss": 0.0426,
+      "step": 3506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028687525234998224,
+      "loss": 0.0291,
+      "step": 3507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002868673211774742,
+      "loss": 0.0082,
+      "step": 3508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002868593877190236,
+      "loss": 0.0155,
+      "step": 3509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028685145197476287,
+      "loss": 0.0198,
+      "step": 3510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028684351394482453,
+      "loss": 0.0131,
+      "step": 3511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002868355736293412,
+      "loss": 0.017,
+      "step": 3512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028682763102844556,
+      "loss": 0.0284,
+      "step": 3513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002868196861422701,
+      "loss": 0.0348,
+      "step": 3514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002868117389709477,
+      "loss": 0.0199,
+      "step": 3515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000286803789514611,
+      "loss": 0.0174,
+      "step": 3516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028679583777339277,
+      "loss": 0.0401,
+      "step": 3517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028678788374742576,
+      "loss": 0.0112,
+      "step": 3518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000286779927436843,
+      "loss": 0.0136,
+      "step": 3519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867719688417772,
+      "loss": 0.014,
+      "step": 3520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028676400796236136,
+      "loss": 0.0337,
+      "step": 3521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867560447987284,
+      "loss": 0.0189,
+      "step": 3522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028674807935101136,
+      "loss": 0.0047,
+      "step": 3523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867401116193433,
+      "loss": 0.0152,
+      "step": 3524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867321416038572,
+      "loss": 0.0314,
+      "step": 3525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867241693046863,
+      "loss": 0.0139,
+      "step": 3526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867161947219636,
+      "loss": 0.0215,
+      "step": 3527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028670821785582246,
+      "loss": 0.0325,
+      "step": 3528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028670023870639604,
+      "loss": 0.0211,
+      "step": 3529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028669225727381755,
+      "loss": 0.0253,
+      "step": 3530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028668427355822034,
+      "loss": 0.0134,
+      "step": 3531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866762875597377,
+      "loss": 0.0178,
+      "step": 3532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028666829927850314,
+      "loss": 0.014,
+      "step": 3533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028666030871465003,
+      "loss": 0.0152,
+      "step": 3534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866523158683117,
+      "loss": 0.0209,
+      "step": 3535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866443207396218,
+      "loss": 0.0497,
+      "step": 3536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028663632332871384,
+      "loss": 0.0384,
+      "step": 3537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028662832363572134,
+      "loss": 0.0285,
+      "step": 3538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028662032166077784,
+      "loss": 0.0169,
+      "step": 3539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866123174040172,
+      "loss": 0.0034,
+      "step": 3540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028660431086557296,
+      "loss": 0.0388,
+      "step": 3541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028659630204557883,
+      "loss": 0.0159,
+      "step": 3542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865882909441686,
+      "loss": 0.0268,
+      "step": 3543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865802775614761,
+      "loss": 0.0534,
+      "step": 3544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865722618976352,
+      "loss": 0.0176,
+      "step": 3545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028656424395277975,
+      "loss": 0.0117,
+      "step": 3546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865562237270435,
+      "loss": 0.021,
+      "step": 3547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028654820122056064,
+      "loss": 0.0173,
+      "step": 3548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865401764334651,
+      "loss": 0.0189,
+      "step": 3549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028653214936589084,
+      "loss": 0.0144,
+      "step": 3550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028652412001797193,
+      "loss": 0.0222,
+      "step": 3551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028651608838984254,
+      "loss": 0.052,
+      "step": 3552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865080544816368,
+      "loss": 0.0288,
+      "step": 3553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865000182934889,
+      "loss": 0.0194,
+      "step": 3554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028649197982553293,
+      "loss": 0.0394,
+      "step": 3555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864839390779033,
+      "loss": 0.0102,
+      "step": 3556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864758960507342,
+      "loss": 0.0294,
+      "step": 3557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864678507441601,
+      "loss": 0.0459,
+      "step": 3558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028645980315831533,
+      "loss": 0.0176,
+      "step": 3559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864517532933342,
+      "loss": 0.042,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002864437011493512,
+      "loss": 0.0099,
+      "step": 3561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028643564672650083,
+      "loss": 0.028,
+      "step": 3562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028642759002491766,
+      "loss": 0.0322,
+      "step": 3563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002864195310447362,
+      "loss": 0.036,
+      "step": 3564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028641146978609103,
+      "loss": 0.0388,
+      "step": 3565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028640340624911683,
+      "loss": 0.0118,
+      "step": 3566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863953404339483,
+      "loss": 0.0123,
+      "step": 3567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863872723407201,
+      "loss": 0.0121,
+      "step": 3568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286379201969567,
+      "loss": 0.021,
+      "step": 3569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863711293206238,
+      "loss": 0.0459,
+      "step": 3570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863630543940253,
+      "loss": 0.0385,
+      "step": 3571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863549771899064,
+      "loss": 0.0169,
+      "step": 3572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286346897708402,
+      "loss": 0.009,
+      "step": 3573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028633881594964706,
+      "loss": 0.0336,
+      "step": 3574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028633073191377655,
+      "loss": 0.0496,
+      "step": 3575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863226456009254,
+      "loss": 0.0175,
+      "step": 3576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863145570112288,
+      "loss": 0.0109,
+      "step": 3577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028630646614482175,
+      "loss": 0.0205,
+      "step": 3578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862983730018394,
+      "loss": 0.0183,
+      "step": 3579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286290277582417,
+      "loss": 0.0149,
+      "step": 3580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862821798866897,
+      "loss": 0.0174,
+      "step": 3581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862740799147927,
+      "loss": 0.0107,
+      "step": 3582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862659776668613,
+      "loss": 0.009,
+      "step": 3583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028625787314303096,
+      "loss": 0.0284,
+      "step": 3584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028624976634343687,
+      "loss": 0.0253,
+      "step": 3585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862416572682145,
+      "loss": 0.0184,
+      "step": 3586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028623354591749924,
+      "loss": 0.0368,
+      "step": 3587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028622543229142665,
+      "loss": 0.0063,
+      "step": 3588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028621731639013215,
+      "loss": 0.0094,
+      "step": 3589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862091982137514,
+      "loss": 0.0162,
+      "step": 3590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028620107776241986,
+      "loss": 0.0221,
+      "step": 3591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861929550362733,
+      "loss": 0.0089,
+      "step": 3592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028618483003544726,
+      "loss": 0.0488,
+      "step": 3593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861767027600775,
+      "loss": 0.0599,
+      "step": 3594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028616857321029976,
+      "loss": 0.0232,
+      "step": 3595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861604413862498,
+      "loss": 0.0382,
+      "step": 3596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028615230728806345,
+      "loss": 0.0961,
+      "step": 3597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028614417091587656,
+      "loss": 0.0846,
+      "step": 3598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028613603226982504,
+      "loss": 0.0154,
+      "step": 3599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861278913500448,
+      "loss": 0.022,
+      "step": 3600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861197481566718,
+      "loss": 0.0665,
+      "step": 3601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028611160268984207,
+      "loss": 0.0128,
+      "step": 3602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028610345494969164,
+      "loss": 0.0342,
+      "step": 3603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028609530493635665,
+      "loss": 0.0401,
+      "step": 3604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002860871526499731,
+      "loss": 0.0418,
+      "step": 3605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028607899809067727,
+      "loss": 0.0148,
+      "step": 3606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002860708412586053,
+      "loss": 0.02,
+      "step": 3607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002860626821538934,
+      "loss": 0.01,
+      "step": 3608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028605452077667785,
+      "loss": 0.0196,
+      "step": 3609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000286046357127095,
+      "loss": 0.0526,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002860381912052812,
+      "loss": 0.0233,
+      "step": 3611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002860300230113728,
+      "loss": 0.0459,
+      "step": 3612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028602185254550627,
+      "loss": 0.0463,
+      "step": 3613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028601367980781806,
+      "loss": 0.0534,
+      "step": 3614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002860055047984446,
+      "loss": 0.0292,
+      "step": 3615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028599732751752245,
+      "loss": 0.028,
+      "step": 3616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859891479651883,
+      "loss": 0.0352,
+      "step": 3617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028598096614157866,
+      "loss": 0.0526,
+      "step": 3618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859727820468302,
+      "loss": 0.0236,
+      "step": 3619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859645956810796,
+      "loss": 0.0175,
+      "step": 3620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859564070444636,
+      "loss": 0.0358,
+      "step": 3621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028594821613711895,
+      "loss": 0.0064,
+      "step": 3622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859400229591825,
+      "loss": 0.0137,
+      "step": 3623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000285931827510791,
+      "loss": 0.0404,
+      "step": 3624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859236297920815,
+      "loss": 0.0131,
+      "step": 3625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028591542980319073,
+      "loss": 0.0285,
+      "step": 3626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028590722754425575,
+      "loss": 0.0133,
+      "step": 3627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858990230154135,
+      "loss": 0.0141,
+      "step": 3628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858908162168011,
+      "loss": 0.0268,
+      "step": 3629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858826071485555,
+      "loss": 0.0078,
+      "step": 3630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858743958108139,
+      "loss": 0.055,
+      "step": 3631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028586618220371334,
+      "loss": 0.0169,
+      "step": 3632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028585796632739114,
+      "loss": 0.0441,
+      "step": 3633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028584974818198443,
+      "loss": 0.0097,
+      "step": 3634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858415277676305,
+      "loss": 0.0233,
+      "step": 3635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858333050844666,
+      "loss": 0.0096,
+      "step": 3636
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028582508013263015,
+      "loss": 0.0212,
+      "step": 3637
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028581685291225844,
+      "loss": 0.0149,
+      "step": 3638
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028580862342348896,
+      "loss": 0.0096,
+      "step": 3639
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858003916664591,
+      "loss": 0.0374,
+      "step": 3640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857921576413063,
+      "loss": 0.0322,
+      "step": 3641
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028578392134816823,
+      "loss": 0.0156,
+      "step": 3642
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857756827871823,
+      "loss": 0.0235,
+      "step": 3643
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857674419584862,
+      "loss": 0.0376,
+      "step": 3644
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857591988622175,
+      "loss": 0.0261,
+      "step": 3645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028575095349851396,
+      "loss": 0.0173,
+      "step": 3646
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857427058675132,
+      "loss": 0.03,
+      "step": 3647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857344559693531,
+      "loss": 0.0074,
+      "step": 3648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028572620380417133,
+      "loss": 0.0423,
+      "step": 3649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857179493721057,
+      "loss": 0.0162,
+      "step": 3650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857096926732942,
+      "loss": 0.0137,
+      "step": 3651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028570143370787465,
+      "loss": 0.0192,
+      "step": 3652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028569317247598494,
+      "loss": 0.0117,
+      "step": 3653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028568490897776315,
+      "loss": 0.0183,
+      "step": 3654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028567664321334734,
+      "loss": 0.025,
+      "step": 3655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028566837518287537,
+      "loss": 0.014,
+      "step": 3656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002856601048864855,
+      "loss": 0.0211,
+      "step": 3657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028565183232431574,
+      "loss": 0.0219,
+      "step": 3658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856435574965044,
+      "loss": 0.0156,
+      "step": 3659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856352804031895,
+      "loss": 0.0042,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028562700104450947,
+      "loss": 0.0177,
+      "step": 3661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856187194206025,
+      "loss": 0.0168,
+      "step": 3662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856104355316069,
+      "loss": 0.0183,
+      "step": 3663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856021493776611,
+      "loss": 0.0432,
+      "step": 3664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028559386095890335,
+      "loss": 0.0163,
+      "step": 3665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028558557027547223,
+      "loss": 0.0083,
+      "step": 3666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028557727732750614,
+      "loss": 0.0117,
+      "step": 3667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855689821151436,
+      "loss": 0.01,
+      "step": 3668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855606846385231,
+      "loss": 0.0353,
+      "step": 3669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028555238489778336,
+      "loss": 0.0194,
+      "step": 3670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855440828930629,
+      "loss": 0.0287,
+      "step": 3671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028553577862450033,
+      "loss": 0.0236,
+      "step": 3672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855274720922345,
+      "loss": 0.0114,
+      "step": 3673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000285519163296404,
+      "loss": 0.0481,
+      "step": 3674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855108522371477,
+      "loss": 0.0203,
+      "step": 3675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855025389146043,
+      "loss": 0.009,
+      "step": 3676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028549422332891283,
+      "loss": 0.0256,
+      "step": 3677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000285485905480212,
+      "loss": 0.0399,
+      "step": 3678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028547758536864084,
+      "loss": 0.0319,
+      "step": 3679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854692629943382,
+      "loss": 0.0357,
+      "step": 3680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028546093835744326,
+      "loss": 0.0166,
+      "step": 3681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854526114580949,
+      "loss": 0.0401,
+      "step": 3682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854442822964322,
+      "loss": 0.0169,
+      "step": 3683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854359508725943,
+      "loss": 0.0346,
+      "step": 3684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854276171867205,
+      "loss": 0.0354,
+      "step": 3685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028541928123894974,
+      "loss": 0.0119,
+      "step": 3686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854109430294214,
+      "loss": 0.0191,
+      "step": 3687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854026025582747,
+      "loss": 0.0585,
+      "step": 3688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028539425982564896,
+      "loss": 0.0073,
+      "step": 3689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853859148316835,
+      "loss": 0.0291,
+      "step": 3690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853775675765177,
+      "loss": 0.014,
+      "step": 3691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028536921806029096,
+      "loss": 0.055,
+      "step": 3692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028536086628314273,
+      "loss": 0.0142,
+      "step": 3693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028535251224521256,
+      "loss": 0.0373,
+      "step": 3694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028534415594663997,
+      "loss": 0.0133,
+      "step": 3695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028533579738756446,
+      "loss": 0.0572,
+      "step": 3696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853274365681256,
+      "loss": 0.0078,
+      "step": 3697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853190734884632,
+      "loss": 0.0665,
+      "step": 3698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028531070814871685,
+      "loss": 0.0099,
+      "step": 3699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853023405490262,
+      "loss": 0.0201,
+      "step": 3700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852939706895311,
+      "loss": 0.0363,
+      "step": 3701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028528559857037125,
+      "loss": 0.004,
+      "step": 3702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028527722419168655,
+      "loss": 0.0101,
+      "step": 3703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852688475536169,
+      "loss": 0.0579,
+      "step": 3704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852604686563021,
+      "loss": 0.0454,
+      "step": 3705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852520874998822,
+      "loss": 0.0382,
+      "step": 3706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852437040844971,
+      "loss": 0.0454,
+      "step": 3707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028523531841028685,
+      "loss": 0.0254,
+      "step": 3708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002852269304773915,
+      "loss": 0.0248,
+      "step": 3709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002852185402859512,
+      "loss": 0.0164,
+      "step": 3710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000285210147836106,
+      "loss": 0.0102,
+      "step": 3711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002852017531279961,
+      "loss": 0.0137,
+      "step": 3712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851933561617617,
+      "loss": 0.0067,
+      "step": 3713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851849569375431,
+      "loss": 0.021,
+      "step": 3714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028517655545548054,
+      "loss": 0.015,
+      "step": 3715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851681517157143,
+      "loss": 0.017,
+      "step": 3716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851597457183848,
+      "loss": 0.014,
+      "step": 3717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851513374636324,
+      "loss": 0.0258,
+      "step": 3718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028514292695159757,
+      "loss": 0.0212,
+      "step": 3719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028513451418242075,
+      "loss": 0.0066,
+      "step": 3720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028512609915624243,
+      "loss": 0.0315,
+      "step": 3721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028511768187320323,
+      "loss": 0.0262,
+      "step": 3722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851092623334436,
+      "loss": 0.0283,
+      "step": 3723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028510084053710443,
+      "loss": 0.0358,
+      "step": 3724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028509241648432606,
+      "loss": 0.0135,
+      "step": 3725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028508399017524935,
+      "loss": 0.0312,
+      "step": 3726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028507556161001503,
+      "loss": 0.0113,
+      "step": 3727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850671307887639,
+      "loss": 0.0199,
+      "step": 3728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028505869771163666,
+      "loss": 0.0181,
+      "step": 3729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028505026237877427,
+      "loss": 0.0322,
+      "step": 3730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850418247903175,
+      "loss": 0.025,
+      "step": 3731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028503338494640745,
+      "loss": 0.029,
+      "step": 3732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850249428471849,
+      "loss": 0.006,
+      "step": 3733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028501649849279096,
+      "loss": 0.0151,
+      "step": 3734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028500805188336664,
+      "loss": 0.027,
+      "step": 3735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000284999603019053,
+      "loss": 0.0164,
+      "step": 3736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028499115189999113,
+      "loss": 0.0318,
+      "step": 3737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849826985263223,
+      "loss": 0.032,
+      "step": 3738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849742428981875,
+      "loss": 0.0194,
+      "step": 3739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028496578501572813,
+      "loss": 0.0167,
+      "step": 3740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849573248790854,
+      "loss": 0.0071,
+      "step": 3741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849488624884005,
+      "loss": 0.0155,
+      "step": 3742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028494039784381493,
+      "loss": 0.0361,
+      "step": 3743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028493193094547,
+      "loss": 0.0578,
+      "step": 3744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849234617935071,
+      "loss": 0.0114,
+      "step": 3745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028491499038806766,
+      "loss": 0.0217,
+      "step": 3746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849065167292933,
+      "loss": 0.0506,
+      "step": 3747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848980408173254,
+      "loss": 0.0277,
+      "step": 3748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028488956265230564,
+      "loss": 0.0386,
+      "step": 3749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848810822343755,
+      "loss": 0.0071,
+      "step": 3750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848725995636767,
+      "loss": 0.0421,
+      "step": 3751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848641146403509,
+      "loss": 0.024,
+      "step": 3752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848556274645398,
+      "loss": 0.0091,
+      "step": 3753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028484713803638513,
+      "loss": 0.0241,
+      "step": 3754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848386463560288,
+      "loss": 0.0176,
+      "step": 3755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848301524236125,
+      "loss": 0.0231,
+      "step": 3756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028482165623927813,
+      "loss": 0.0313,
+      "step": 3757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028481315780316764,
+      "loss": 0.0239,
+      "step": 3758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028480465711542294,
+      "loss": 0.0538,
+      "step": 3759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028479615417618595,
+      "loss": 0.0427,
+      "step": 3760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847876489855988,
+      "loss": 0.0302,
+      "step": 3761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028477914154380346,
+      "loss": 0.0088,
+      "step": 3762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000284770631850942,
+      "loss": 0.0424,
+      "step": 3763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028476211990715667,
+      "loss": 0.018,
+      "step": 3764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847536057125895,
+      "loss": 0.0079,
+      "step": 3765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028474508926738275,
+      "loss": 0.047,
+      "step": 3766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028473657057167867,
+      "loss": 0.0094,
+      "step": 3767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847280496256196,
+      "loss": 0.0191,
+      "step": 3768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028471952642934766,
+      "loss": 0.0315,
+      "step": 3769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847110009830054,
+      "loss": 0.0312,
+      "step": 3770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847024732867351,
+      "loss": 0.0167,
+      "step": 3771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028469394334067926,
+      "loss": 0.0091,
+      "step": 3772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846854111449803,
+      "loss": 0.0237,
+      "step": 3773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028467687669978073,
+      "loss": 0.0171,
+      "step": 3774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028466834000522316,
+      "loss": 0.0114,
+      "step": 3775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028465980106145,
+      "loss": 0.0106,
+      "step": 3776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028465125986860403,
+      "loss": 0.0325,
+      "step": 3777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846427164268279,
+      "loss": 0.0193,
+      "step": 3778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028463417073626413,
+      "loss": 0.0146,
+      "step": 3779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846256227970557,
+      "loss": 0.041,
+      "step": 3780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028461707260934516,
+      "loss": 0.0185,
+      "step": 3781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028460852017327544,
+      "loss": 0.0186,
+      "step": 3782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028459996548898934,
+      "loss": 0.0194,
+      "step": 3783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845914085566297,
+      "loss": 0.015,
+      "step": 3784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028458284937633947,
+      "loss": 0.0308,
+      "step": 3785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028457428794826165,
+      "loss": 0.0177,
+      "step": 3786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845657242725392,
+      "loss": 0.0311,
+      "step": 3787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028455715834931517,
+      "loss": 0.0251,
+      "step": 3788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028454859017873257,
+      "loss": 0.0042,
+      "step": 3789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845400197609345,
+      "loss": 0.0235,
+      "step": 3790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845314470960642,
+      "loss": 0.0121,
+      "step": 3791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845228721842647,
+      "loss": 0.0368,
+      "step": 3792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845142950256794,
+      "loss": 0.0433,
+      "step": 3793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845057156204514,
+      "loss": 0.0305,
+      "step": 3794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028449713396872406,
+      "loss": 0.0351,
+      "step": 3795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028448855007064073,
+      "loss": 0.0144,
+      "step": 3796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028447996392634475,
+      "loss": 0.0214,
+      "step": 3797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844713755359795,
+      "loss": 0.0193,
+      "step": 3798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844627848996884,
+      "loss": 0.0108,
+      "step": 3799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028445419201761504,
+      "loss": 0.0247,
+      "step": 3800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844455968899029,
+      "loss": 0.0118,
+      "step": 3801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028443699951669543,
+      "loss": 0.0086,
+      "step": 3802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028442839989813634,
+      "loss": 0.0441,
+      "step": 3803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844197980343692,
+      "loss": 0.0471,
+      "step": 3804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844111939255377,
+      "loss": 0.0208,
+      "step": 3805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844025875717856,
+      "loss": 0.0156,
+      "step": 3806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028439397897325654,
+      "loss": 0.0208,
+      "step": 3807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028438536813009434,
+      "loss": 0.0212,
+      "step": 3808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843767550424428,
+      "loss": 0.0294,
+      "step": 3809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843681397104458,
+      "loss": 0.0242,
+      "step": 3810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843595221342473,
+      "loss": 0.0237,
+      "step": 3811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843509023139911,
+      "loss": 0.0019,
+      "step": 3812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028434228024982127,
+      "loss": 0.0077,
+      "step": 3813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843336559418817,
+      "loss": 0.0179,
+      "step": 3814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843250293903166,
+      "loss": 0.0154,
+      "step": 3815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028431640059526985,
+      "loss": 0.0267,
+      "step": 3816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028430776955688575,
+      "loss": 0.0152,
+      "step": 3817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842991362753083,
+      "loss": 0.029,
+      "step": 3818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028429050075068184,
+      "loss": 0.012,
+      "step": 3819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028428186298315043,
+      "loss": 0.0301,
+      "step": 3820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028427322297285853,
+      "loss": 0.038,
+      "step": 3821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842645807199503,
+      "loss": 0.0148,
+      "step": 3822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028425593622457017,
+      "loss": 0.0302,
+      "step": 3823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028424728948686244,
+      "loss": 0.0342,
+      "step": 3824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028423864050697154,
+      "loss": 0.0119,
+      "step": 3825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000284229989285042,
+      "loss": 0.0319,
+      "step": 3826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842213358212183,
+      "loss": 0.0189,
+      "step": 3827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028421268011564484,
+      "loss": 0.0124,
+      "step": 3828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842040221684663,
+      "loss": 0.0459,
+      "step": 3829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028419536197982735,
+      "loss": 0.0686,
+      "step": 3830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841866995498725,
+      "loss": 0.0168,
+      "step": 3831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841780348787464,
+      "loss": 0.0107,
+      "step": 3832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028416936796659385,
+      "loss": 0.0169,
+      "step": 3833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028416069881355963,
+      "loss": 0.0647,
+      "step": 3834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841520274197885,
+      "loss": 0.0187,
+      "step": 3835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841433537854253,
+      "loss": 0.0211,
+      "step": 3836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841346779106148,
+      "loss": 0.0242,
+      "step": 3837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028412599979550207,
+      "loss": 0.0109,
+      "step": 3838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841173194402319,
+      "loss": 0.0183,
+      "step": 3839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028410863684494937,
+      "loss": 0.0312,
+      "step": 3840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028409995200979945,
+      "loss": 0.0044,
+      "step": 3841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840912649349272,
+      "loss": 0.0076,
+      "step": 3842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028408257562047767,
+      "loss": 0.0202,
+      "step": 3843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028407388406659606,
+      "loss": 0.0297,
+      "step": 3844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840651902734275,
+      "loss": 0.0333,
+      "step": 3845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840564942411172,
+      "loss": 0.0343,
+      "step": 3846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840477959698104,
+      "loss": 0.0343,
+      "step": 3847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028403909545965237,
+      "loss": 0.0182,
+      "step": 3848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840303927107885,
+      "loss": 0.0156,
+      "step": 3849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028402168772336396,
+      "loss": 0.014,
+      "step": 3850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840129804975243,
+      "loss": 0.0339,
+      "step": 3851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028400427103341487,
+      "loss": 0.0136,
+      "step": 3852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002839955593311812,
+      "loss": 0.0333,
+      "step": 3853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002839868453909687,
+      "loss": 0.016,
+      "step": 3854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000283978129212923,
+      "loss": 0.0296,
+      "step": 3855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839694107971897,
+      "loss": 0.0495,
+      "step": 3856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839606901439143,
+      "loss": 0.0364,
+      "step": 3857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028395196725324246,
+      "loss": 0.0252,
+      "step": 3858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028394324212531994,
+      "loss": 0.0235,
+      "step": 3859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839345147602925,
+      "loss": 0.0148,
+      "step": 3860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839257851583058,
+      "loss": 0.0601,
+      "step": 3861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028391705331950567,
+      "loss": 0.018,
+      "step": 3862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028390831924403803,
+      "loss": 0.0096,
+      "step": 3863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838995829320486,
+      "loss": 0.0223,
+      "step": 3864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028389084438368344,
+      "loss": 0.007,
+      "step": 3865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838821035990884,
+      "loss": 0.0246,
+      "step": 3866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028387336057840955,
+      "loss": 0.0177,
+      "step": 3867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028386461532179286,
+      "loss": 0.0194,
+      "step": 3868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028385586782938443,
+      "loss": 0.0456,
+      "step": 3869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838471181013303,
+      "loss": 0.023,
+      "step": 3870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028383836613777663,
+      "loss": 0.0178,
+      "step": 3871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838296119388697,
+      "loss": 0.0177,
+      "step": 3872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838208555047555,
+      "loss": 0.0363,
+      "step": 3873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028381209683558045,
+      "loss": 0.011,
+      "step": 3874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028380333593149086,
+      "loss": 0.0126,
+      "step": 3875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837945727926329,
+      "loss": 0.0217,
+      "step": 3876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028378580741915307,
+      "loss": 0.0294,
+      "step": 3877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837770398111977,
+      "loss": 0.0121,
+      "step": 3878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028376826996891327,
+      "loss": 0.0207,
+      "step": 3879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028375949789244626,
+      "loss": 0.0771,
+      "step": 3880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028375072358194305,
+      "loss": 0.056,
+      "step": 3881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028374194703755033,
+      "loss": 0.028,
+      "step": 3882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028373316825941464,
+      "loss": 0.0595,
+      "step": 3883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837243872476826,
+      "loss": 0.0251,
+      "step": 3884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837156040025009,
+      "loss": 0.0249,
+      "step": 3885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837068185240161,
+      "loss": 0.0181,
+      "step": 3886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028369803081237515,
+      "loss": 0.0381,
+      "step": 3887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836892408677247,
+      "loss": 0.0183,
+      "step": 3888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836804486902115,
+      "loss": 0.016,
+      "step": 3889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028367165427998256,
+      "loss": 0.0061,
+      "step": 3890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836628576371846,
+      "loss": 0.0107,
+      "step": 3891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028365405876196463,
+      "loss": 0.0092,
+      "step": 3892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028364525765446966,
+      "loss": 0.0166,
+      "step": 3893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836364543148465,
+      "loss": 0.0247,
+      "step": 3894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028362764874324234,
+      "loss": 0.0336,
+      "step": 3895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028361884093980425,
+      "loss": 0.016,
+      "step": 3896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028361003090467926,
+      "loss": 0.0203,
+      "step": 3897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836012186380145,
+      "loss": 0.0217,
+      "step": 3898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002835924041399573,
+      "loss": 0.0116,
+      "step": 3899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002835835874106547,
+      "loss": 0.0176,
+      "step": 3900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028357476845025404,
+      "loss": 0.0213,
+      "step": 3901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002835659472589026,
+      "loss": 0.0081,
+      "step": 3902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028355712383674775,
+      "loss": 0.0207,
+      "step": 3903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002835482981839368,
+      "loss": 0.0156,
+      "step": 3904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002835394703006172,
+      "loss": 0.0365,
+      "step": 3905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002835306401869364,
+      "loss": 0.0306,
+      "step": 3906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028352180784304177,
+      "loss": 0.0098,
+      "step": 3907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283512973269081,
+      "loss": 0.0178,
+      "step": 3908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002835041364652015,
+      "loss": 0.0138,
+      "step": 3909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834952974315509,
+      "loss": 0.0098,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834864561682769,
+      "loss": 0.0046,
+      "step": 3911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028347761267552703,
+      "loss": 0.0062,
+      "step": 3912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028346876695344914,
+      "loss": 0.0243,
+      "step": 3913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028345991900219087,
+      "loss": 0.0232,
+      "step": 3914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834510688219,
+      "loss": 0.009,
+      "step": 3915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834422164127244,
+      "loss": 0.0126,
+      "step": 3916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834333617748119,
+      "loss": 0.0193,
+      "step": 3917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834245049083104,
+      "loss": 0.0172,
+      "step": 3918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834156458133678,
+      "loss": 0.0158,
+      "step": 3919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834067844901321,
+      "loss": 0.0429,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028339792093875127,
+      "loss": 0.0275,
+      "step": 3921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833890551593733,
+      "loss": 0.0149,
+      "step": 3922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833801871521464,
+      "loss": 0.0372,
+      "step": 3923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833713169172185,
+      "loss": 0.0112,
+      "step": 3924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028336244445473794,
+      "loss": 0.0141,
+      "step": 3925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833535697648528,
+      "loss": 0.0231,
+      "step": 3926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833446928477113,
+      "loss": 0.0533,
+      "step": 3927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833358137034617,
+      "loss": 0.0268,
+      "step": 3928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833269323322524,
+      "loss": 0.0105,
+      "step": 3929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028331804873423156,
+      "loss": 0.0509,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833091629095477,
+      "loss": 0.0082,
+      "step": 3931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028330027485834924,
+      "loss": 0.0213,
+      "step": 3932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832913845807845,
+      "loss": 0.0162,
+      "step": 3933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283282492077002,
+      "loss": 0.0242,
+      "step": 3934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832735973471504,
+      "loss": 0.0166,
+      "step": 3935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028326470039137806,
+      "loss": 0.0252,
+      "step": 3936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832558012098337,
+      "loss": 0.019,
+      "step": 3937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832468998026659,
+      "loss": 0.0165,
+      "step": 3938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832379961700234,
+      "loss": 0.0095,
+      "step": 3939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028322909031205484,
+      "loss": 0.0238,
+      "step": 3940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283220182228909,
+      "loss": 0.0097,
+      "step": 3941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832112719207346,
+      "loss": 0.026,
+      "step": 3942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832023593876805,
+      "loss": 0.0118,
+      "step": 3943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028319344462989556,
+      "loss": 0.0248,
+      "step": 3944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831845276475287,
+      "loss": 0.0185,
+      "step": 3945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831756084407289,
+      "loss": 0.0154,
+      "step": 3946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831666870096449,
+      "loss": 0.0236,
+      "step": 3947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283157763354426,
+      "loss": 0.0055,
+      "step": 3948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283148837475221,
+      "loss": 0.0217,
+      "step": 3949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028313990937217914,
+      "loss": 0.0155,
+      "step": 3950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831309790454494,
+      "loss": 0.0234,
+      "step": 3951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831220464951811,
+      "loss": 0.0293,
+      "step": 3952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028311311172152324,
+      "loss": 0.0183,
+      "step": 3953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028310417472462524,
+      "loss": 0.019,
+      "step": 3954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830952355046362,
+      "loss": 0.022,
+      "step": 3955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028308629406170553,
+      "loss": 0.015,
+      "step": 3956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028307735039598253,
+      "loss": 0.0065,
+      "step": 3957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028306840450761657,
+      "loss": 0.0464,
+      "step": 3958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028305945639675703,
+      "loss": 0.02,
+      "step": 3959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028305050606355344,
+      "loss": 0.022,
+      "step": 3960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830415535081552,
+      "loss": 0.0116,
+      "step": 3961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000283032598730712,
+      "loss": 0.0216,
+      "step": 3962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028302364173137315,
+      "loss": 0.0363,
+      "step": 3963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830146825102885,
+      "loss": 0.0088,
+      "step": 3964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830057210676075,
+      "loss": 0.0128,
+      "step": 3965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028299675740347987,
+      "loss": 0.0123,
+      "step": 3966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829877915180554,
+      "loss": 0.0168,
+      "step": 3967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028297882341148377,
+      "loss": 0.0146,
+      "step": 3968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028296985308391476,
+      "loss": 0.0316,
+      "step": 3969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829608805354982,
+      "loss": 0.013,
+      "step": 3970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028295190576638397,
+      "loss": 0.0211,
+      "step": 3971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028294292877672193,
+      "loss": 0.0162,
+      "step": 3972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028293394956666203,
+      "loss": 0.0044,
+      "step": 3973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829249681363543,
+      "loss": 0.047,
+      "step": 3974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829159844859486,
+      "loss": 0.0164,
+      "step": 3975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028290699861559514,
+      "loss": 0.0094,
+      "step": 3976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828980105254439,
+      "loss": 0.0345,
+      "step": 3977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000282889020215645,
+      "loss": 0.0599,
+      "step": 3978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828800276863486,
+      "loss": 0.0208,
+      "step": 3979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028287103293770494,
+      "loss": 0.0109,
+      "step": 3980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028286203596986423,
+      "loss": 0.0079,
+      "step": 3981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828530367829767,
+      "loss": 0.0231,
+      "step": 3982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828440353771927,
+      "loss": 0.0047,
+      "step": 3983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828350317526625,
+      "loss": 0.0188,
+      "step": 3984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828260259095366,
+      "loss": 0.0132,
+      "step": 3985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828170178479653,
+      "loss": 0.0254,
+      "step": 3986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828080075680991,
+      "loss": 0.0224,
+      "step": 3987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028279899507008846,
+      "loss": 0.0259,
+      "step": 3988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827899803540839,
+      "loss": 0.0248,
+      "step": 3989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028278096342023606,
+      "loss": 0.0272,
+      "step": 3990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827719442686955,
+      "loss": 0.0315,
+      "step": 3991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028276292289961284,
+      "loss": 0.0229,
+      "step": 3992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827538993131387,
+      "loss": 0.0164,
+      "step": 3993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827448735094239,
+      "loss": 0.0598,
+      "step": 3994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028273584548861916,
+      "loss": 0.0118,
+      "step": 3995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827268152508752,
+      "loss": 0.0422,
+      "step": 3996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827177827963429,
+      "loss": 0.0367,
+      "step": 3997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827087481251731,
+      "loss": 0.0108,
+      "step": 3998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002826997112375167,
+      "loss": 0.025,
+      "step": 3999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028269067213352466,
+      "loss": 0.0224,
+      "step": 4000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002826816308133479,
+      "loss": 0.0262,
+      "step": 4001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028267258727713744,
+      "loss": 0.0303,
+      "step": 4002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028266354152504437,
+      "loss": 0.0164,
+      "step": 4003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028265449355721975,
+      "loss": 0.0348,
+      "step": 4004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002826454433738146,
+      "loss": 0.0277,
+      "step": 4005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028263639097498026,
+      "loss": 0.0136,
+      "step": 4006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028262733636086774,
+      "loss": 0.0141,
+      "step": 4007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028261827953162836,
+      "loss": 0.0206,
+      "step": 4008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002826092204874134,
+      "loss": 0.0193,
+      "step": 4009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002826001592283742,
+      "loss": 0.0303,
+      "step": 4010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028259109575466196,
+      "loss": 0.0089,
+      "step": 4011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028258203006642816,
+      "loss": 0.0243,
+      "step": 4012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825729621638242,
+      "loss": 0.0028,
+      "step": 4013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028256389204700153,
+      "loss": 0.0152,
+      "step": 4014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825548197161116,
+      "loss": 0.0072,
+      "step": 4015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028254574517130594,
+      "loss": 0.0387,
+      "step": 4016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825366684127362,
+      "loss": 0.0193,
+      "step": 4017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825275894405539,
+      "loss": 0.0204,
+      "step": 4018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825185082549107,
+      "loss": 0.0214,
+      "step": 4019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028250942485595826,
+      "loss": 0.0261,
+      "step": 4020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028250033924384827,
+      "loss": 0.0147,
+      "step": 4021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028249125141873257,
+      "loss": 0.0209,
+      "step": 4022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824821613807628,
+      "loss": 0.0154,
+      "step": 4023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028247306913009086,
+      "loss": 0.015,
+      "step": 4024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028246397466686864,
+      "loss": 0.0202,
+      "step": 4025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028245487799124797,
+      "loss": 0.0097,
+      "step": 4026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028244577910338086,
+      "loss": 0.0133,
+      "step": 4027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824366780034192,
+      "loss": 0.0084,
+      "step": 4028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000282427574691515,
+      "loss": 0.0218,
+      "step": 4029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028241846916782036,
+      "loss": 0.0849,
+      "step": 4030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824093614324873,
+      "loss": 0.0154,
+      "step": 4031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000282400251485668,
+      "loss": 0.0192,
+      "step": 4032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823911393275146,
+      "loss": 0.0203,
+      "step": 4033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823820249581792,
+      "loss": 0.0586,
+      "step": 4034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028237290837781413,
+      "loss": 0.025,
+      "step": 4035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823637895865716,
+      "loss": 0.0154,
+      "step": 4036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000282354668584604,
+      "loss": 0.0294,
+      "step": 4037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823455453720635,
+      "loss": 0.0245,
+      "step": 4038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028233641994910257,
+      "loss": 0.012,
+      "step": 4039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028232729231587373,
+      "loss": 0.0278,
+      "step": 4040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028231816247252924,
+      "loss": 0.013,
+      "step": 4041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823090304192217,
+      "loss": 0.0124,
+      "step": 4042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822998961561036,
+      "loss": 0.0749,
+      "step": 4043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028229075968332746,
+      "loss": 0.006,
+      "step": 4044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028228162100104595,
+      "loss": 0.0252,
+      "step": 4045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028227248010941167,
+      "loss": 0.0256,
+      "step": 4046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822633370085773,
+      "loss": 0.0183,
+      "step": 4047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822541916986955,
+      "loss": 0.0164,
+      "step": 4048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822450441799191,
+      "loss": 0.0151,
+      "step": 4049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822358944524008,
+      "loss": 0.0232,
+      "step": 4050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822267425162934,
+      "loss": 0.0103,
+      "step": 4051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002822175883717499,
+      "loss": 0.0248,
+      "step": 4052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028220843201892307,
+      "loss": 0.0259,
+      "step": 4053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821992734579658,
+      "loss": 0.0105,
+      "step": 4054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821901126890312,
+      "loss": 0.0118,
+      "step": 4055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821809497122722,
+      "loss": 0.011,
+      "step": 4056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821717845278418,
+      "loss": 0.0274,
+      "step": 4057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821626171358931,
+      "loss": 0.0375,
+      "step": 4058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028215344753657913,
+      "loss": 0.0124,
+      "step": 4059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821442757300533,
+      "loss": 0.0397,
+      "step": 4060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821351017164685,
+      "loss": 0.0285,
+      "step": 4061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821259254959781,
+      "loss": 0.0192,
+      "step": 4062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821167470687354,
+      "loss": 0.0253,
+      "step": 4063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821075664348936,
+      "loss": 0.0236,
+      "step": 4064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028209838359460605,
+      "loss": 0.008,
+      "step": 4065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820891985480262,
+      "loss": 0.009,
+      "step": 4066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820800112953073,
+      "loss": 0.0191,
+      "step": 4067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028207082183660294,
+      "loss": 0.0194,
+      "step": 4068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028206163017206656,
+      "loss": 0.0419,
+      "step": 4069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028205243630185166,
+      "loss": 0.0108,
+      "step": 4070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028204324022611184,
+      "loss": 0.028,
+      "step": 4071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028203404194500066,
+      "loss": 0.0276,
+      "step": 4072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820248414586717,
+      "loss": 0.0179,
+      "step": 4073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028201563876727875,
+      "loss": 0.0097,
+      "step": 4074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028200643387097534,
+      "loss": 0.0131,
+      "step": 4075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819972267699154,
+      "loss": 0.0013,
+      "step": 4076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028198801746425254,
+      "loss": 0.0017,
+      "step": 4077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819788059541406,
+      "loss": 0.0253,
+      "step": 4078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028196959223973357,
+      "loss": 0.0383,
+      "step": 4079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819603763211852,
+      "loss": 0.0091,
+      "step": 4080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819511581986495,
+      "loss": 0.0197,
+      "step": 4081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819419378722803,
+      "loss": 0.0379,
+      "step": 4082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819327153422317,
+      "loss": 0.0231,
+      "step": 4083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028192349060865776,
+      "loss": 0.012,
+      "step": 4084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819142636717125,
+      "loss": 0.0042,
+      "step": 4085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028190503453155,
+      "loss": 0.0295,
+      "step": 4086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028189580318832444,
+      "loss": 0.0058,
+      "step": 4087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028188656964219,
+      "loss": 0.011,
+      "step": 4088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818773338933009,
+      "loss": 0.0173,
+      "step": 4089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028186809594181136,
+      "loss": 0.0157,
+      "step": 4090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028185885578787574,
+      "loss": 0.0124,
+      "step": 4091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818496134316483,
+      "loss": 0.0315,
+      "step": 4092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818403688732834,
+      "loss": 0.0194,
+      "step": 4093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028183112211293556,
+      "loss": 0.0153,
+      "step": 4094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000281821873150759,
+      "loss": 0.0209,
+      "step": 4095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818126219869085,
+      "loss": 0.01,
+      "step": 4096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818033686215383,
+      "loss": 0.0196,
+      "step": 4097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000281794113054803,
+      "loss": 0.0138,
+      "step": 4098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002817848552868573,
+      "loss": 0.0178,
+      "step": 4099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002817755953178558,
+      "loss": 0.0165,
+      "step": 4100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028176633314795304,
+      "loss": 0.0182,
+      "step": 4101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817570687773038,
+      "loss": 0.0247,
+      "step": 4102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817478022060628,
+      "loss": 0.0212,
+      "step": 4103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028173853343438484,
+      "loss": 0.0373,
+      "step": 4104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028172926246242466,
+      "loss": 0.0205,
+      "step": 4105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028171998929033714,
+      "loss": 0.0444,
+      "step": 4106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817107139182772,
+      "loss": 0.0444,
+      "step": 4107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817014363463997,
+      "loss": 0.0249,
+      "step": 4108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028169215657485963,
+      "loss": 0.0049,
+      "step": 4109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816828746038119,
+      "loss": 0.0219,
+      "step": 4110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028167359043341166,
+      "loss": 0.008,
+      "step": 4111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816643040638139,
+      "loss": 0.0312,
+      "step": 4112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816550154951737,
+      "loss": 0.0211,
+      "step": 4113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816457247276462,
+      "loss": 0.0267,
+      "step": 4114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816364317613866,
+      "loss": 0.0213,
+      "step": 4115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816271365965502,
+      "loss": 0.0055,
+      "step": 4116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028161783923329203,
+      "loss": 0.0214,
+      "step": 4117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816085396717676,
+      "loss": 0.0419,
+      "step": 4118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028159923791213205,
+      "loss": 0.0207,
+      "step": 4119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815899339545409,
+      "loss": 0.0343,
+      "step": 4120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815806277991494,
+      "loss": 0.0211,
+      "step": 4121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028157131944611305,
+      "loss": 0.0167,
+      "step": 4122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815620088955873,
+      "loss": 0.0325,
+      "step": 4123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815526961477277,
+      "loss": 0.0426,
+      "step": 4124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028154338120268973,
+      "loss": 0.0224,
+      "step": 4125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000281534064060629,
+      "loss": 0.0326,
+      "step": 4126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028152474472170104,
+      "loss": 0.0361,
+      "step": 4127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028151542318606164,
+      "loss": 0.031,
+      "step": 4128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815060994538664,
+      "loss": 0.0058,
+      "step": 4129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814967735252711,
+      "loss": 0.032,
+      "step": 4130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814874454004314,
+      "loss": 0.0211,
+      "step": 4131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814781150795032,
+      "loss": 0.0348,
+      "step": 4132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814687825626423,
+      "loss": 0.0245,
+      "step": 4133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028145944785000455,
+      "loss": 0.0394,
+      "step": 4134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814501109417459,
+      "loss": 0.0169,
+      "step": 4135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814407718380222,
+      "loss": 0.0191,
+      "step": 4136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028143143053898956,
+      "loss": 0.0145,
+      "step": 4137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028142208704480396,
+      "loss": 0.0202,
+      "step": 4138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028141274135562135,
+      "loss": 0.0491,
+      "step": 4139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028140339347159797,
+      "loss": 0.023,
+      "step": 4140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813940433928898,
+      "loss": 0.0255,
+      "step": 4141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813846911196532,
+      "loss": 0.0652,
+      "step": 4142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028137533665204413,
+      "loss": 0.0176,
+      "step": 4143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000281365979990219,
+      "loss": 0.011,
+      "step": 4144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028135662113433404,
+      "loss": 0.0282,
+      "step": 4145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813472600845455,
+      "loss": 0.0226,
+      "step": 4146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028133789684100987,
+      "loss": 0.1191,
+      "step": 4147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813285314038834,
+      "loss": 0.0241,
+      "step": 4148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813191637733226,
+      "loss": 0.0443,
+      "step": 4149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002813097939494838,
+      "loss": 0.0242,
+      "step": 4150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028130042193252365,
+      "loss": 0.012,
+      "step": 4151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812910477225986,
+      "loss": 0.0224,
+      "step": 4152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028128167131986513,
+      "loss": 0.0124,
+      "step": 4153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028127229272448,
+      "loss": 0.0239,
+      "step": 4154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812629119365998,
+      "loss": 0.0267,
+      "step": 4155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812535289563812,
+      "loss": 0.0115,
+      "step": 4156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812441437839809,
+      "loss": 0.0168,
+      "step": 4157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812347564195557,
+      "loss": 0.0273,
+      "step": 4158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812253668632622,
+      "loss": 0.0286,
+      "step": 4159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812159751152575,
+      "loss": 0.0176,
+      "step": 4160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812065811756983,
+      "loss": 0.0216,
+      "step": 4161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028119718504474153,
+      "loss": 0.0372,
+      "step": 4162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811877867225441,
+      "loss": 0.0206,
+      "step": 4163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000281178386209263,
+      "loss": 0.0267,
+      "step": 4164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028116898350505527,
+      "loss": 0.0209,
+      "step": 4165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811595786100779,
+      "loss": 0.017,
+      "step": 4166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028115017152448795,
+      "loss": 0.0129,
+      "step": 4167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028114076224844257,
+      "loss": 0.0303,
+      "step": 4168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028113135078209895,
+      "loss": 0.0267,
+      "step": 4169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028112193712561426,
+      "loss": 0.0457,
+      "step": 4170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028111252127914564,
+      "loss": 0.0199,
+      "step": 4171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028110310324285046,
+      "loss": 0.006,
+      "step": 4172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000281093683016886,
+      "loss": 0.027,
+      "step": 4173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028108426060140956,
+      "loss": 0.0149,
+      "step": 4174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810748359965785,
+      "loss": 0.0076,
+      "step": 4175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028106540920255036,
+      "loss": 0.009,
+      "step": 4176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810559802194824,
+      "loss": 0.0138,
+      "step": 4177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810465490475321,
+      "loss": 0.0192,
+      "step": 4178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810371156868572,
+      "loss": 0.0302,
+      "step": 4179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810276801376151,
+      "loss": 0.0129,
+      "step": 4180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028101824239996334,
+      "loss": 0.0178,
+      "step": 4181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810088024740597,
+      "loss": 0.0333,
+      "step": 4182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809993603600617,
+      "loss": 0.0293,
+      "step": 4183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028098991605812713,
+      "loss": 0.014,
+      "step": 4184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028098046956841373,
+      "loss": 0.0117,
+      "step": 4185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809710208910792,
+      "loss": 0.013,
+      "step": 4186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809615700262814,
+      "loss": 0.0203,
+      "step": 4187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809521169741782,
+      "loss": 0.0781,
+      "step": 4188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028094266173492744,
+      "loss": 0.0175,
+      "step": 4189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809332043086871,
+      "loss": 0.0124,
+      "step": 4190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000280923744695615,
+      "loss": 0.0224,
+      "step": 4191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028091428289586933,
+      "loss": 0.0277,
+      "step": 4192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809048189096079,
+      "loss": 0.0393,
+      "step": 4193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000280895352736989,
+      "loss": 0.0142,
+      "step": 4194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028088588437817055,
+      "loss": 0.0035,
+      "step": 4195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002808764138333108,
+      "loss": 0.0271,
+      "step": 4196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028086694110256795,
+      "loss": 0.051,
+      "step": 4197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028085746618610006,
+      "loss": 0.0264,
+      "step": 4198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002808479890840655,
+      "loss": 0.081,
+      "step": 4199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028083850979662256,
+      "loss": 0.0251,
+      "step": 4200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028082902832392954,
+      "loss": 0.017,
+      "step": 4201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028081954466614473,
+      "loss": 0.0162,
+      "step": 4202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028081005882342664,
+      "loss": 0.0195,
+      "step": 4203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002808005707959336,
+      "loss": 0.0197,
+      "step": 4204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807910805838242,
+      "loss": 0.0288,
+      "step": 4205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807815881872568,
+      "loss": 0.0332,
+      "step": 4206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028077209360639,
+      "loss": 0.0135,
+      "step": 4207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807625968413824,
+      "loss": 0.0212,
+      "step": 4208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028075309789239267,
+      "loss": 0.0149,
+      "step": 4209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807435967595793,
+      "loss": 0.0107,
+      "step": 4210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807340934431011,
+      "loss": 0.0278,
+      "step": 4211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028072458794311676,
+      "loss": 0.0136,
+      "step": 4212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028071508025978505,
+      "loss": 0.015,
+      "step": 4213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028070557039326476,
+      "loss": 0.0112,
+      "step": 4214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002806960583437147,
+      "loss": 0.0313,
+      "step": 4215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028068654411129377,
+      "loss": 0.0169,
+      "step": 4216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028067702769616086,
+      "loss": 0.0234,
+      "step": 4217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002806675090984749,
+      "loss": 0.0179,
+      "step": 4218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028065798831839497,
+      "loss": 0.0096,
+      "step": 4219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002806484653560799,
+      "loss": 0.0089,
+      "step": 4220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028063894021168893,
+      "loss": 0.0366,
+      "step": 4221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000280629412885381,
+      "loss": 0.0379,
+      "step": 4222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028061988337731527,
+      "loss": 0.0228,
+      "step": 4223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000280610351687651,
+      "loss": 0.0071,
+      "step": 4224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002806008178165473,
+      "loss": 0.0221,
+      "step": 4225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028059128176416334,
+      "loss": 0.0501,
+      "step": 4226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028058174353065855,
+      "loss": 0.0268,
+      "step": 4227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028057220311619214,
+      "loss": 0.02,
+      "step": 4228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805626605209234,
+      "loss": 0.0252,
+      "step": 4229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028055311574501184,
+      "loss": 0.0271,
+      "step": 4230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805435687886168,
+      "loss": 0.0074,
+      "step": 4231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805340196518977,
+      "loss": 0.0249,
+      "step": 4232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805244683350141,
+      "loss": 0.0099,
+      "step": 4233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028051491483812555,
+      "loss": 0.0091,
+      "step": 4234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805053591613915,
+      "loss": 0.0271,
+      "step": 4235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804958013049716,
+      "loss": 0.0076,
+      "step": 4236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804862412690255,
+      "loss": 0.0207,
+      "step": 4237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804766790537129,
+      "loss": 0.049,
+      "step": 4238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804671146591934,
+      "loss": 0.0363,
+      "step": 4239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028045754808562687,
+      "loss": 0.0218,
+      "step": 4240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000280447979333173,
+      "loss": 0.0089,
+      "step": 4241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028043840840199163,
+      "loss": 0.0372,
+      "step": 4242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028042883529224265,
+      "loss": 0.0115,
+      "step": 4243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804192600040859,
+      "loss": 0.0194,
+      "step": 4244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804096825376813,
+      "loss": 0.0163,
+      "step": 4245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028040010289318885,
+      "loss": 0.0211,
+      "step": 4246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028039052107076855,
+      "loss": 0.0236,
+      "step": 4247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002803809370705804,
+      "loss": 0.0244,
+      "step": 4248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002803713508927845,
+      "loss": 0.0309,
+      "step": 4249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280361762537541,
+      "loss": 0.0413,
+      "step": 4250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028035217200500994,
+      "loss": 0.0411,
+      "step": 4251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028034257929535154,
+      "loss": 0.0181,
+      "step": 4252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028033298440872603,
+      "loss": 0.0099,
+      "step": 4253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028032338734529366,
+      "loss": 0.0179,
+      "step": 4254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028031378810521474,
+      "loss": 0.0274,
+      "step": 4255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028030418668864956,
+      "loss": 0.0242,
+      "step": 4256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028029458309575847,
+      "loss": 0.0143,
+      "step": 4257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028028497732670194,
+      "loss": 0.046,
+      "step": 4258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802753693816403,
+      "loss": 0.0166,
+      "step": 4259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802657592607341,
+      "loss": 0.0369,
+      "step": 4260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028025614696414387,
+      "loss": 0.0302,
+      "step": 4261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028024653249203007,
+      "loss": 0.0247,
+      "step": 4262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028023691584455336,
+      "loss": 0.014,
+      "step": 4263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802272970218743,
+      "loss": 0.0321,
+      "step": 4264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802176760241535,
+      "loss": 0.0628,
+      "step": 4265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802080528515518,
+      "loss": 0.0024,
+      "step": 4266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801984275042298,
+      "loss": 0.0259,
+      "step": 4267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801887999823482,
+      "loss": 0.0036,
+      "step": 4268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280179170286068,
+      "loss": 0.0059,
+      "step": 4269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801695384155499,
+      "loss": 0.0054,
+      "step": 4270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801599043709548,
+      "loss": 0.023,
+      "step": 4271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801502681524436,
+      "loss": 0.0287,
+      "step": 4272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028014062976017723,
+      "loss": 0.0082,
+      "step": 4273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801309891943167,
+      "loss": 0.0307,
+      "step": 4274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280121346455023,
+      "loss": 0.0417,
+      "step": 4275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801117015424572,
+      "loss": 0.0122,
+      "step": 4276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801020544567804,
+      "loss": 0.0077,
+      "step": 4277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800924051981537,
+      "loss": 0.0418,
+      "step": 4278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028008275376673823,
+      "loss": 0.0423,
+      "step": 4279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800731001626953,
+      "loss": 0.0217,
+      "step": 4280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280063444386186,
+      "loss": 0.0542,
+      "step": 4281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800537864373716,
+      "loss": 0.0103,
+      "step": 4282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800441263164136,
+      "loss": 0.0717,
+      "step": 4283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800344640234732,
+      "loss": 0.0219,
+      "step": 4284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800247995587117,
+      "loss": 0.0061,
+      "step": 4285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800151329222907,
+      "loss": 0.0123,
+      "step": 4286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800054641143715,
+      "loss": 0.0372,
+      "step": 4287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027999579313511564,
+      "loss": 0.02,
+      "step": 4288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027998611998468476,
+      "loss": 0.0194,
+      "step": 4289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799764446632402,
+      "loss": 0.0457,
+      "step": 4290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027996676717094367,
+      "loss": 0.03,
+      "step": 4291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027995708750795683,
+      "loss": 0.0322,
+      "step": 4292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799474056744413,
+      "loss": 0.0174,
+      "step": 4293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799377216705589,
+      "loss": 0.0253,
+      "step": 4294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027992803549647114,
+      "loss": 0.0389,
+      "step": 4295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027991834715234,
+      "loss": 0.0044,
+      "step": 4296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799086566383272,
+      "loss": 0.0212,
+      "step": 4297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027989896395459464,
+      "loss": 0.0133,
+      "step": 4298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027988926910130414,
+      "loss": 0.0304,
+      "step": 4299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798795720786177,
+      "loss": 0.011,
+      "step": 4300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798698728866972,
+      "loss": 0.0131,
+      "step": 4301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798601715257047,
+      "loss": 0.0443,
+      "step": 4302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027985046799580227,
+      "loss": 0.0157,
+      "step": 4303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798407622971518,
+      "loss": 0.0377,
+      "step": 4304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027983105442991557,
+      "loss": 0.0067,
+      "step": 4305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027982134439425566,
+      "loss": 0.0148,
+      "step": 4306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027981163219033424,
+      "loss": 0.0119,
+      "step": 4307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027980191781831347,
+      "loss": 0.0279,
+      "step": 4308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027979220127835575,
+      "loss": 0.0261,
+      "step": 4309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027978248257062323,
+      "loss": 0.0209,
+      "step": 4310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797727616952782,
+      "loss": 0.034,
+      "step": 4311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027976303865248316,
+      "loss": 0.033,
+      "step": 4312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027975331344240044,
+      "loss": 0.0266,
+      "step": 4313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797435860651924,
+      "loss": 0.0198,
+      "step": 4314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797338565210216,
+      "loss": 0.01,
+      "step": 4315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027972412481005055,
+      "loss": 0.0147,
+      "step": 4316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027971439093244166,
+      "loss": 0.0249,
+      "step": 4317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027970465488835766,
+      "loss": 0.0226,
+      "step": 4318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027969491667796106,
+      "loss": 0.0392,
+      "step": 4319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796851763014145,
+      "loss": 0.0215,
+      "step": 4320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796754337588807,
+      "loss": 0.027,
+      "step": 4321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027966568905052244,
+      "loss": 0.0152,
+      "step": 4322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027965594217650236,
+      "loss": 0.021,
+      "step": 4323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027964619313698333,
+      "loss": 0.0106,
+      "step": 4324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027963644193212814,
+      "loss": 0.0159,
+      "step": 4325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027962668856209964,
+      "loss": 0.0098,
+      "step": 4326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796169330270608,
+      "loss": 0.015,
+      "step": 4327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027960717532717446,
+      "loss": 0.0293,
+      "step": 4328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027959741546260364,
+      "loss": 0.0616,
+      "step": 4329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795876534335114,
+      "loss": 0.0168,
+      "step": 4330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027957788924006075,
+      "loss": 0.0165,
+      "step": 4331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027956812288241464,
+      "loss": 0.0233,
+      "step": 4332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795583543607364,
+      "loss": 0.0109,
+      "step": 4333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795485836751891,
+      "loss": 0.0184,
+      "step": 4334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795388108259358,
+      "loss": 0.0398,
+      "step": 4335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027952903581313993,
+      "loss": 0.0244,
+      "step": 4336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027951925863696465,
+      "loss": 0.012,
+      "step": 4337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027950947929757325,
+      "loss": 0.0229,
+      "step": 4338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794996977951291,
+      "loss": 0.0255,
+      "step": 4339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027948991412979547,
+      "loss": 0.0178,
+      "step": 4340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027948012830173596,
+      "loss": 0.0193,
+      "step": 4341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794703403111138,
+      "loss": 0.0205,
+      "step": 4342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794605501580926,
+      "loss": 0.0141,
+      "step": 4343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027945075784283584,
+      "loss": 0.0203,
+      "step": 4344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794409633655071,
+      "loss": 0.0246,
+      "step": 4345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794311667262699,
+      "loss": 0.0162,
+      "step": 4346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002794213679252879,
+      "loss": 0.0956,
+      "step": 4347
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027941156696272473,
+      "loss": 0.0221,
+      "step": 4348
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002794017638387442,
+      "loss": 0.0212,
+      "step": 4349
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793919585535099,
+      "loss": 0.0348,
+      "step": 4350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793821511071856,
+      "loss": 0.0221,
+      "step": 4351
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793723414999352,
+      "loss": 0.0068,
+      "step": 4352
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027936252973192246,
+      "loss": 0.0098,
+      "step": 4353
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793527158033113,
+      "loss": 0.0065,
+      "step": 4354
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793428997142656,
+      "loss": 0.0197,
+      "step": 4355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793330814649493,
+      "loss": 0.0418,
+      "step": 4356
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793232610555265,
+      "loss": 0.02,
+      "step": 4357
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027931343848616106,
+      "loss": 0.0152,
+      "step": 4358
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793036137570171,
+      "loss": 0.0211,
+      "step": 4359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792937868682587,
+      "loss": 0.0073,
+      "step": 4360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027928395782005007,
+      "loss": 0.0122,
+      "step": 4361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792741266125553,
+      "loss": 0.0126,
+      "step": 4362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792642932459385,
+      "loss": 0.0273,
+      "step": 4363
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027925445772036405,
+      "loss": 0.0097,
+      "step": 4364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792446200359962,
+      "loss": 0.0211,
+      "step": 4365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792347801929992,
+      "loss": 0.055,
+      "step": 4366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792249381915374,
+      "loss": 0.0314,
+      "step": 4367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792150940317752,
+      "loss": 0.01,
+      "step": 4368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027920524771387706,
+      "loss": 0.0241,
+      "step": 4369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027919539923800733,
+      "loss": 0.0129,
+      "step": 4370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027918554860433063,
+      "loss": 0.035,
+      "step": 4371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027917569581301137,
+      "loss": 0.035,
+      "step": 4372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002791658408642142,
+      "loss": 0.0572,
+      "step": 4373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027915598375810363,
+      "loss": 0.0179,
+      "step": 4374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027914612449484437,
+      "loss": 0.0293,
+      "step": 4375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027913626307460104,
+      "loss": 0.0139,
+      "step": 4376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002791263994975383,
+      "loss": 0.011,
+      "step": 4377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000279116533763821,
+      "loss": 0.0374,
+      "step": 4378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027910666587361384,
+      "loss": 0.0228,
+      "step": 4379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027909679582708164,
+      "loss": 0.0273,
+      "step": 4380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027908692362438927,
+      "loss": 0.0135,
+      "step": 4381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790770492657016,
+      "loss": 0.0116,
+      "step": 4382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790671727511836,
+      "loss": 0.0388,
+      "step": 4383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027905729408100015,
+      "loss": 0.0385,
+      "step": 4384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790474132553163,
+      "loss": 0.0267,
+      "step": 4385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000279037530274297,
+      "loss": 0.0051,
+      "step": 4386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027902764513810743,
+      "loss": 0.0281,
+      "step": 4387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790177578469126,
+      "loss": 0.0311,
+      "step": 4388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790078684008777,
+      "loss": 0.0287,
+      "step": 4389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027899797680016784,
+      "loss": 0.021,
+      "step": 4390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002789880830449483,
+      "loss": 0.0853,
+      "step": 4391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027897818713538427,
+      "loss": 0.021,
+      "step": 4392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000278968289071641,
+      "loss": 0.0052,
+      "step": 4393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000278958388853884,
+      "loss": 0.0196,
+      "step": 4394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027894848648227834,
+      "loss": 0.0101,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002789385819569896,
+      "loss": 0.0099,
+      "step": 4396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002789286752781832,
+      "loss": 0.0166,
+      "step": 4397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002789187664460245,
+      "loss": 0.0429,
+      "step": 4398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027890885546067904,
+      "loss": 0.0144,
+      "step": 4399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788989423223124,
+      "loss": 0.0109,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788890270310901,
+      "loss": 0.043,
+      "step": 4401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788791095871777,
+      "loss": 0.0557,
+      "step": 4402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027886918999074097,
+      "loss": 0.0194,
+      "step": 4403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027885926824194543,
+      "loss": 0.0256,
+      "step": 4404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788493443409569,
+      "loss": 0.0251,
+      "step": 4405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788394182879412,
+      "loss": 0.0021,
+      "step": 4406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788294900830639,
+      "loss": 0.0323,
+      "step": 4407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027881955972649093,
+      "loss": 0.0068,
+      "step": 4408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788096272183882,
+      "loss": 0.0097,
+      "step": 4409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787996925589215,
+      "loss": 0.0115,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787897557482569,
+      "loss": 0.0194,
+      "step": 4411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027877981678656015,
+      "loss": 0.0058,
+      "step": 4412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027876987567399745,
+      "loss": 0.0216,
+      "step": 4413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027875993241073473,
+      "loss": 0.0081,
+      "step": 4414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787499869969381,
+      "loss": 0.0062,
+      "step": 4415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027874003943277365,
+      "loss": 0.0177,
+      "step": 4416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787300897184075,
+      "loss": 0.0222,
+      "step": 4417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027872013785400587,
+      "loss": 0.0937,
+      "step": 4418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278710183839735,
+      "loss": 0.0638,
+      "step": 4419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027870022767576105,
+      "loss": 0.0145,
+      "step": 4420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786902693622504,
+      "loss": 0.0113,
+      "step": 4421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027868030889936925,
+      "loss": 0.0103,
+      "step": 4422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786703462872842,
+      "loss": 0.0084,
+      "step": 4423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786603815261613,
+      "loss": 0.0265,
+      "step": 4424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027865041461616727,
+      "loss": 0.0221,
+      "step": 4425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786404455574685,
+      "loss": 0.0076,
+      "step": 4426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786304743502314,
+      "loss": 0.0603,
+      "step": 4427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027862050099462256,
+      "loss": 0.0286,
+      "step": 4428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786105254908086,
+      "loss": 0.0102,
+      "step": 4429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786005478389561,
+      "loss": 0.0166,
+      "step": 4430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785905680392317,
+      "loss": 0.0087,
+      "step": 4431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027858058609180205,
+      "loss": 0.0108,
+      "step": 4432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785706019968339,
+      "loss": 0.0646,
+      "step": 4433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027856061575449407,
+      "loss": 0.0145,
+      "step": 4434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785506273649492,
+      "loss": 0.0373,
+      "step": 4435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027854063682836627,
+      "loss": 0.012,
+      "step": 4436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027853064414491203,
+      "loss": 0.0449,
+      "step": 4437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027852064931475337,
+      "loss": 0.0249,
+      "step": 4438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785106523380573,
+      "loss": 0.0052,
+      "step": 4439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027850065321499077,
+      "loss": 0.0433,
+      "step": 4440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027849065194572077,
+      "loss": 0.0385,
+      "step": 4441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002784806485304143,
+      "loss": 0.0164,
+      "step": 4442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002784706429692385,
+      "loss": 0.0101,
+      "step": 4443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002784606352623605,
+      "loss": 0.0162,
+      "step": 4444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002784506254099474,
+      "loss": 0.04,
+      "step": 4445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027844061341216635,
+      "loss": 0.0279,
+      "step": 4446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027843059926918454,
+      "loss": 0.0365,
+      "step": 4447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027842058298116945,
+      "loss": 0.0285,
+      "step": 4448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002784105645482881,
+      "loss": 0.0172,
+      "step": 4449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000278400543970708,
+      "loss": 0.0193,
+      "step": 4450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027839052124859637,
+      "loss": 0.0165,
+      "step": 4451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783804963821207,
+      "loss": 0.0337,
+      "step": 4452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783704693714484,
+      "loss": 0.0424,
+      "step": 4453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027836044021674696,
+      "loss": 0.0306,
+      "step": 4454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783504089181839,
+      "loss": 0.0152,
+      "step": 4455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783403754759267,
+      "loss": 0.0556,
+      "step": 4456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027833033989014297,
+      "loss": 0.0271,
+      "step": 4457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783203021610003,
+      "loss": 0.0207,
+      "step": 4458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783102622886664,
+      "loss": 0.0124,
+      "step": 4459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783002202733089,
+      "loss": 0.0283,
+      "step": 4460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027829017611509554,
+      "loss": 0.0179,
+      "step": 4461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782801298141941,
+      "loss": 0.0057,
+      "step": 4462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782700813707723,
+      "loss": 0.0131,
+      "step": 4463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027826003078499804,
+      "loss": 0.0493,
+      "step": 4464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027824997805703917,
+      "loss": 0.0481,
+      "step": 4465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782399231870636,
+      "loss": 0.021,
+      "step": 4466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782298661752392,
+      "loss": 0.0188,
+      "step": 4467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027821980702173396,
+      "loss": 0.0262,
+      "step": 4468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027820974572671597,
+      "loss": 0.0149,
+      "step": 4469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781996822903532,
+      "loss": 0.0121,
+      "step": 4470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781896167128137,
+      "loss": 0.0348,
+      "step": 4471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781795489942656,
+      "loss": 0.035,
+      "step": 4472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027816947913487716,
+      "loss": 0.0303,
+      "step": 4473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781594071348164,
+      "loss": 0.0171,
+      "step": 4474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027814933299425174,
+      "loss": 0.0268,
+      "step": 4475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027813925671335126,
+      "loss": 0.0114,
+      "step": 4476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027812917829228334,
+      "loss": 0.025,
+      "step": 4477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781190977312162,
+      "loss": 0.0241,
+      "step": 4478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781090150303183,
+      "loss": 0.0336,
+      "step": 4479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027809893018975807,
+      "loss": 0.0142,
+      "step": 4480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780888432097039,
+      "loss": 0.005,
+      "step": 4481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027807875409032427,
+      "loss": 0.0259,
+      "step": 4482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027806866283178763,
+      "loss": 0.0377,
+      "step": 4483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780585694342627,
+      "loss": 0.0074,
+      "step": 4484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780484738979178,
+      "loss": 0.0172,
+      "step": 4485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780383762229217,
+      "loss": 0.011,
+      "step": 4486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027802827640944307,
+      "loss": 0.0127,
+      "step": 4487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780181744576505,
+      "loss": 0.025,
+      "step": 4488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780080703677128,
+      "loss": 0.016,
+      "step": 4489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002779979641397987,
+      "loss": 0.0146,
+      "step": 4490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000277987855774077,
+      "loss": 0.0158,
+      "step": 4491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027797774527071645,
+      "loss": 0.0198,
+      "step": 4492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000277967632629886,
+      "loss": 0.0166,
+      "step": 4493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027795751785175454,
+      "loss": 0.0104,
+      "step": 4494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027794740093649105,
+      "loss": 0.0244,
+      "step": 4495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002779372818842644,
+      "loss": 0.0302,
+      "step": 4496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027792716069524365,
+      "loss": 0.0323,
+      "step": 4497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002779170373695978,
+      "loss": 0.015,
+      "step": 4498
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000277906911907496,
+      "loss": 0.0178,
+      "step": 4499
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778967843091073,
+      "loss": 0.0138,
+      "step": 4500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027788665457460093,
+      "loss": 0.0153,
+      "step": 4501
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000277876522704146,
+      "loss": 0.0122,
+      "step": 4502
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027786638869791173,
+      "loss": 0.0114,
+      "step": 4503
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027785625255606747,
+      "loss": 0.0491,
+      "step": 4504
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027784611427878234,
+      "loss": 0.0269,
+      "step": 4505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778359738662259,
+      "loss": 0.0197,
+      "step": 4506
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778258313185673,
+      "loss": 0.0157,
+      "step": 4507
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027781568663597605,
+      "loss": 0.0361,
+      "step": 4508
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778055398186216,
+      "loss": 0.0097,
+      "step": 4509
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027779539086667334,
+      "loss": 0.003,
+      "step": 4510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777852397803008,
+      "loss": 0.014,
+      "step": 4511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777750865596736,
+      "loss": 0.0348,
+      "step": 4512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027776493120496124,
+      "loss": 0.0081,
+      "step": 4513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027775477371633336,
+      "loss": 0.0743,
+      "step": 4514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777446140939596,
+      "loss": 0.0329,
+      "step": 4515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777344523380097,
+      "loss": 0.014,
+      "step": 4516
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777242884486533,
+      "loss": 0.0141,
+      "step": 4517
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027771412242606015,
+      "loss": 0.0084,
+      "step": 4518
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777039542704001,
+      "loss": 0.0094,
+      "step": 4519
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027769378398184296,
+      "loss": 0.0187,
+      "step": 4520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776836115605586,
+      "loss": 0.0284,
+      "step": 4521
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776734370067169,
+      "loss": 0.0395,
+      "step": 4522
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027766326032048776,
+      "loss": 0.0372,
+      "step": 4523
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776530815020412,
+      "loss": 0.0852,
+      "step": 4524
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027764290055154727,
+      "loss": 0.0159,
+      "step": 4525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776327174691759,
+      "loss": 0.0075,
+      "step": 4526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027762253225509725,
+      "loss": 0.0199,
+      "step": 4527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776123449094814,
+      "loss": 0.0251,
+      "step": 4528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776021554324985,
+      "loss": 0.0191,
+      "step": 4529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027759196382431874,
+      "loss": 0.0201,
+      "step": 4530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775817700851124,
+      "loss": 0.0254,
+      "step": 4531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775715742150496,
+      "loss": 0.0223,
+      "step": 4532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775613762143007,
+      "loss": 0.0306,
+      "step": 4533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775511760830361,
+      "loss": 0.0162,
+      "step": 4534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000277540973821426,
+      "loss": 0.0066,
+      "step": 4535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027753076942964094,
+      "loss": 0.0172,
+      "step": 4536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027752056290785126,
+      "loss": 0.0381,
+      "step": 4537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775103542562275,
+      "loss": 0.0139,
+      "step": 4538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775001434749401,
+      "loss": 0.0127,
+      "step": 4539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002774899305641597,
+      "loss": 0.0112,
+      "step": 4540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027747971552405674,
+      "loss": 0.0262,
+      "step": 4541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002774694983548019,
+      "loss": 0.0304,
+      "step": 4542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027745927905656586,
+      "loss": 0.0311,
+      "step": 4543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774490576295192,
+      "loss": 0.031,
+      "step": 4544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774388340738328,
+      "loss": 0.0237,
+      "step": 4545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774286083896772,
+      "loss": 0.0748,
+      "step": 4546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027741838057722336,
+      "loss": 0.0187,
+      "step": 4547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774081506366421,
+      "loss": 0.0261,
+      "step": 4548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027739791856810415,
+      "loss": 0.0382,
+      "step": 4549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027738768437178047,
+      "loss": 0.0264,
+      "step": 4550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027737744804784203,
+      "loss": 0.0353,
+      "step": 4551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027736720959645975,
+      "loss": 0.0237,
+      "step": 4552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027735696901780467,
+      "loss": 0.0065,
+      "step": 4553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027734672631204776,
+      "loss": 0.0282,
+      "step": 4554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027733648147936017,
+      "loss": 0.0083,
+      "step": 4555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000277326234519913,
+      "loss": 0.0347,
+      "step": 4556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027731598543387733,
+      "loss": 0.0259,
+      "step": 4557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027730573422142437,
+      "loss": 0.0566,
+      "step": 4558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027729548088272534,
+      "loss": 0.0155,
+      "step": 4559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027728522541795156,
+      "loss": 0.0013,
+      "step": 4560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027727496782727415,
+      "loss": 0.0634,
+      "step": 4561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772647081108646,
+      "loss": 0.0162,
+      "step": 4562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772544462688941,
+      "loss": 0.0696,
+      "step": 4563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027724418230153433,
+      "loss": 0.0283,
+      "step": 4564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027723391620895635,
+      "loss": 0.0165,
+      "step": 4565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027722364799133187,
+      "loss": 0.02,
+      "step": 4566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772133776488323,
+      "loss": 0.0117,
+      "step": 4567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772031051816292,
+      "loss": 0.0362,
+      "step": 4568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771928305898941,
+      "loss": 0.0157,
+      "step": 4569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027718255387379866,
+      "loss": 0.0072,
+      "step": 4570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771722750335145,
+      "loss": 0.0304,
+      "step": 4571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771619940692133,
+      "loss": 0.0132,
+      "step": 4572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771517109810667,
+      "loss": 0.0068,
+      "step": 4573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771414257692466,
+      "loss": 0.0171,
+      "step": 4574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027713113843392466,
+      "loss": 0.019,
+      "step": 4575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027712084897527274,
+      "loss": 0.0215,
+      "step": 4576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027711055739346267,
+      "loss": 0.0486,
+      "step": 4577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027710026368866635,
+      "loss": 0.0101,
+      "step": 4578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770899678610557,
+      "loss": 0.0041,
+      "step": 4579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770796699108027,
+      "loss": 0.045,
+      "step": 4580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027706936983807934,
+      "loss": 0.037,
+      "step": 4581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027705906764305757,
+      "loss": 0.0368,
+      "step": 4582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027704876332590965,
+      "loss": 0.0083,
+      "step": 4583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770384568868075,
+      "loss": 0.0114,
+      "step": 4584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770281483259233,
+      "loss": 0.0087,
+      "step": 4585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770178376434293,
+      "loss": 0.0187,
+      "step": 4586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027700752483949756,
+      "loss": 0.0239,
+      "step": 4587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002769972099143005,
+      "loss": 0.0423,
+      "step": 4588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002769868928680103,
+      "loss": 0.018,
+      "step": 4589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027697657370079923,
+      "loss": 0.0045,
+      "step": 4590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002769662524128397,
+      "loss": 0.0174,
+      "step": 4591
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002769559290043041,
+      "loss": 0.011,
+      "step": 4592
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027694560347536485,
+      "loss": 0.01,
+      "step": 4593
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002769352758261944,
+      "loss": 0.0255,
+      "step": 4594
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027692494605696525,
+      "loss": 0.0343,
+      "step": 4595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027691461416784984,
+      "loss": 0.0099,
+      "step": 4596
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002769042801590209,
+      "loss": 0.0252,
+      "step": 4597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768939440306509,
+      "loss": 0.0294,
+      "step": 4598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027688360578291245,
+      "loss": 0.0571,
+      "step": 4599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027687326541597833,
+      "loss": 0.0535,
+      "step": 4600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768629229300212,
+      "loss": 0.0127,
+      "step": 4601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027685257832521374,
+      "loss": 0.0166,
+      "step": 4602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768422316017288,
+      "loss": 0.0197,
+      "step": 4603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768318827597392,
+      "loss": 0.0163,
+      "step": 4604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027682153179941766,
+      "loss": 0.0212,
+      "step": 4605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768111787209372,
+      "loss": 0.0149,
+      "step": 4606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768008235244707,
+      "loss": 0.0089,
+      "step": 4607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027679046621019107,
+      "loss": 0.0064,
+      "step": 4608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767801067782713,
+      "loss": 0.0288,
+      "step": 4609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767697452288844,
+      "loss": 0.0269,
+      "step": 4610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027675938156220356,
+      "loss": 0.0178,
+      "step": 4611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027674901577840166,
+      "loss": 0.0053,
+      "step": 4612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027673864787765203,
+      "loss": 0.0201,
+      "step": 4613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767282778601277,
+      "loss": 0.0665,
+      "step": 4614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767179057260019,
+      "loss": 0.0529,
+      "step": 4615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027670753147544797,
+      "loss": 0.0321,
+      "step": 4616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000276697155108639,
+      "loss": 0.0123,
+      "step": 4617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027668677662574836,
+      "loss": 0.0241,
+      "step": 4618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027667639602694947,
+      "loss": 0.0042,
+      "step": 4619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027666601331241563,
+      "loss": 0.0173,
+      "step": 4620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002766556284823203,
+      "loss": 0.0566,
+      "step": 4621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027664524153683687,
+      "loss": 0.007,
+      "step": 4622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027663485247613883,
+      "loss": 0.0321,
+      "step": 4623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027662446130039973,
+      "loss": 0.0121,
+      "step": 4624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002766140680097931,
+      "loss": 0.0157,
+      "step": 4625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027660367260449255,
+      "loss": 0.0135,
+      "step": 4626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765932750846717,
+      "loss": 0.0137,
+      "step": 4627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765828754505041,
+      "loss": 0.0369,
+      "step": 4628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765724737021636,
+      "loss": 0.0234,
+      "step": 4629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765620698398239,
+      "loss": 0.0221,
+      "step": 4630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765516638636587,
+      "loss": 0.0143,
+      "step": 4631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765412557738418,
+      "loss": 0.0416,
+      "step": 4632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765308455705471,
+      "loss": 0.0217,
+      "step": 4633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027652043325394844,
+      "loss": 0.0165,
+      "step": 4634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027651001882421974,
+      "loss": 0.0052,
+      "step": 4635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027649960228153485,
+      "loss": 0.0144,
+      "step": 4636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002764891836260679,
+      "loss": 0.0258,
+      "step": 4637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002764787628579928,
+      "loss": 0.0326,
+      "step": 4638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027646833997748354,
+      "loss": 0.0306,
+      "step": 4639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002764579149847143,
+      "loss": 0.0183,
+      "step": 4640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027644748787985927,
+      "loss": 0.0557,
+      "step": 4641
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002764370586630924,
+      "loss": 0.034,
+      "step": 4642
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000276426627334588,
+      "loss": 0.007,
+      "step": 4643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002764161938945203,
+      "loss": 0.0327,
+      "step": 4644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027640575834306355,
+      "loss": 0.0091,
+      "step": 4645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027639532068039204,
+      "loss": 0.027,
+      "step": 4646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027638488090668003,
+      "loss": 0.0179,
+      "step": 4647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000276374439022102,
+      "loss": 0.0372,
+      "step": 4648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763639950268322,
+      "loss": 0.0259,
+      "step": 4649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027635354892104524,
+      "loss": 0.0253,
+      "step": 4650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027634310070491546,
+      "loss": 0.0182,
+      "step": 4651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763326503786174,
+      "loss": 0.0064,
+      "step": 4652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763221979423257,
+      "loss": 0.0147,
+      "step": 4653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027631174339621476,
+      "loss": 0.0318,
+      "step": 4654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763012867404593,
+      "loss": 0.0345,
+      "step": 4655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027629082797523395,
+      "loss": 0.0067,
+      "step": 4656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027628036710071333,
+      "loss": 0.0333,
+      "step": 4657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027626990411707227,
+      "loss": 0.028,
+      "step": 4658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027625943902448544,
+      "loss": 0.0032,
+      "step": 4659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002762489718231276,
+      "loss": 0.0377,
+      "step": 4660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027623850251317364,
+      "loss": 0.0599,
+      "step": 4661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027622803109479845,
+      "loss": 0.0259,
+      "step": 4662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027621755756817684,
+      "loss": 0.0213,
+      "step": 4663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027620708193348375,
+      "loss": 0.0264,
+      "step": 4664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027619660419089413,
+      "loss": 0.0192,
+      "step": 4665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027618612434058307,
+      "loss": 0.0376,
+      "step": 4666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761756423827255,
+      "loss": 0.05,
+      "step": 4667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761651583174966,
+      "loss": 0.0142,
+      "step": 4668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761546721450713,
+      "loss": 0.0233,
+      "step": 4669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761441838656249,
+      "loss": 0.0056,
+      "step": 4670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761336934793325,
+      "loss": 0.0105,
+      "step": 4671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027612320098636934,
+      "loss": 0.0223,
+      "step": 4672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761127063869107,
+      "loss": 0.0185,
+      "step": 4673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761022096811317,
+      "loss": 0.0043,
+      "step": 4674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760917108692078,
+      "loss": 0.0265,
+      "step": 4675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027608120995131427,
+      "loss": 0.0185,
+      "step": 4676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027607070692762663,
+      "loss": 0.0257,
+      "step": 4677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027606020179832014,
+      "loss": 0.0157,
+      "step": 4678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760496945635704,
+      "loss": 0.0069,
+      "step": 4679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027603918522355275,
+      "loss": 0.0104,
+      "step": 4680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027602867377844286,
+      "loss": 0.0238,
+      "step": 4681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760181602284162,
+      "loss": 0.0395,
+      "step": 4682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760076445736483,
+      "loss": 0.0066,
+      "step": 4683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000275997126814315,
+      "loss": 0.024,
+      "step": 4684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027598660695059177,
+      "loss": 0.0146,
+      "step": 4685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759760849826544,
+      "loss": 0.0175,
+      "step": 4686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027596556091067867,
+      "loss": 0.0234,
+      "step": 4687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759550347348402,
+      "loss": 0.023,
+      "step": 4688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000275944506455315,
+      "loss": 0.0512,
+      "step": 4689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027593397607227874,
+      "loss": 0.0128,
+      "step": 4690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002759234435859074,
+      "loss": 0.0111,
+      "step": 4691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002759129089963768,
+      "loss": 0.0153,
+      "step": 4692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275902372303863,
+      "loss": 0.0457,
+      "step": 4693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027589183350854187,
+      "loss": 0.0303,
+      "step": 4694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758812926105895,
+      "loss": 0.0175,
+      "step": 4695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275870749610182,
+      "loss": 0.0372,
+      "step": 4696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027586020450749525,
+      "loss": 0.0177,
+      "step": 4697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758496573027056,
+      "loss": 0.0094,
+      "step": 4698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758391079959891,
+      "loss": 0.0062,
+      "step": 4699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027582855658752194,
+      "loss": 0.0289,
+      "step": 4700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758180030774804,
+      "loss": 0.0117,
+      "step": 4701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027580744746604066,
+      "loss": 0.0522,
+      "step": 4702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757968897533791,
+      "loss": 0.0219,
+      "step": 4703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275786329939672,
+      "loss": 0.0287,
+      "step": 4704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027577576802509575,
+      "loss": 0.0108,
+      "step": 4705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757652040098268,
+      "loss": 0.0285,
+      "step": 4706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027575463789404153,
+      "loss": 0.0244,
+      "step": 4707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757440696779164,
+      "loss": 0.0136,
+      "step": 4708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027573349936162796,
+      "loss": 0.0097,
+      "step": 4709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757229269453527,
+      "loss": 0.0178,
+      "step": 4710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027571235242926736,
+      "loss": 0.0259,
+      "step": 4711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027570177581354833,
+      "loss": 0.0286,
+      "step": 4712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756911970983724,
+      "loss": 0.0132,
+      "step": 4713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756806162839162,
+      "loss": 0.0401,
+      "step": 4714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756700333703565,
+      "loss": 0.016,
+      "step": 4715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027565944835787,
+      "loss": 0.0147,
+      "step": 4716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756488612466335,
+      "loss": 0.0249,
+      "step": 4717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027563827203682393,
+      "loss": 0.0237,
+      "step": 4718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275627680728618,
+      "loss": 0.033,
+      "step": 4719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027561708732219264,
+      "loss": 0.0246,
+      "step": 4720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756064918177248,
+      "loss": 0.0234,
+      "step": 4721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755958942153914,
+      "loss": 0.0192,
+      "step": 4722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027558529451536957,
+      "loss": 0.0111,
+      "step": 4723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755746927178363,
+      "loss": 0.013,
+      "step": 4724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755640888229685,
+      "loss": 0.0278,
+      "step": 4725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755534828309435,
+      "loss": 0.0327,
+      "step": 4726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755428747419383,
+      "loss": 0.0154,
+      "step": 4727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027553226455613014,
+      "loss": 0.0383,
+      "step": 4728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027552165227369614,
+      "loss": 0.0361,
+      "step": 4729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755110378948137,
+      "loss": 0.0089,
+      "step": 4730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755004214196599,
+      "loss": 0.0195,
+      "step": 4731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002754898028484122,
+      "loss": 0.0066,
+      "step": 4732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275479182181248,
+      "loss": 0.0252,
+      "step": 4733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002754685594183445,
+      "loss": 0.0457,
+      "step": 4734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027545793455987925,
+      "loss": 0.0202,
+      "step": 4735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027544730760602963,
+      "loss": 0.013,
+      "step": 4736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027543667855697325,
+      "loss": 0.0191,
+      "step": 4737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027542604741288753,
+      "loss": 0.0261,
+      "step": 4738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027541541417395006,
+      "loss": 0.0139,
+      "step": 4739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002754047788403384,
+      "loss": 0.0066,
+      "step": 4740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753941414122303,
+      "loss": 0.0248,
+      "step": 4741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027538350188980324,
+      "loss": 0.0204,
+      "step": 4742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753728602732351,
+      "loss": 0.0208,
+      "step": 4743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027536221656270345,
+      "loss": 0.0083,
+      "step": 4744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027535157075838626,
+      "loss": 0.0095,
+      "step": 4745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753409228604611,
+      "loss": 0.0523,
+      "step": 4746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000275330272869106,
+      "loss": 0.0456,
+      "step": 4747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753196207844988,
+      "loss": 0.0279,
+      "step": 4748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753089666068173,
+      "loss": 0.0242,
+      "step": 4749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752983103362395,
+      "loss": 0.0248,
+      "step": 4750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752876519729435,
+      "loss": 0.0167,
+      "step": 4751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027527699151710716,
+      "loss": 0.0092,
+      "step": 4752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027526632896890857,
+      "loss": 0.0178,
+      "step": 4753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027525566432852586,
+      "loss": 0.0169,
+      "step": 4754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752449975961371,
+      "loss": 0.0454,
+      "step": 4755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752343287719205,
+      "loss": 0.0309,
+      "step": 4756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027522365785605414,
+      "loss": 0.0189,
+      "step": 4757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027521298484871635,
+      "loss": 0.0156,
+      "step": 4758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752023097500854,
+      "loss": 0.066,
+      "step": 4759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027519163256033955,
+      "loss": 0.0475,
+      "step": 4760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027518095327965707,
+      "loss": 0.0258,
+      "step": 4761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027517027190821636,
+      "loss": 0.005,
+      "step": 4762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002751595884461959,
+      "loss": 0.0071,
+      "step": 4763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000275148902893774,
+      "loss": 0.0222,
+      "step": 4764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027513821525112917,
+      "loss": 0.0217,
+      "step": 4765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027512752551843997,
+      "loss": 0.0427,
+      "step": 4766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002751168336958849,
+      "loss": 0.0104,
+      "step": 4767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002751061397836425,
+      "loss": 0.0317,
+      "step": 4768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027509544378189144,
+      "loss": 0.0425,
+      "step": 4769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002750847456908103,
+      "loss": 0.0157,
+      "step": 4770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002750740455105779,
+      "loss": 0.0223,
+      "step": 4771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027506334324137275,
+      "loss": 0.0399,
+      "step": 4772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002750526388833737,
+      "loss": 0.0134,
+      "step": 4773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027504193243675954,
+      "loss": 0.0355,
+      "step": 4774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027503122390170904,
+      "loss": 0.0072,
+      "step": 4775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002750205132784011,
+      "loss": 0.015,
+      "step": 4776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027500980056701464,
+      "loss": 0.0252,
+      "step": 4777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027499908576772853,
+      "loss": 0.0254,
+      "step": 4778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749883688807217,
+      "loss": 0.0177,
+      "step": 4779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749776499061732,
+      "loss": 0.029,
+      "step": 4780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000274966928844262,
+      "loss": 0.016,
+      "step": 4781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027495620569516725,
+      "loss": 0.0198,
+      "step": 4782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027494548045906795,
+      "loss": 0.0281,
+      "step": 4783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749347531361433,
+      "loss": 0.0171,
+      "step": 4784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749240237265725,
+      "loss": 0.0492,
+      "step": 4785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749132922305346,
+      "loss": 0.0252,
+      "step": 4786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027490255864820896,
+      "loss": 0.0177,
+      "step": 4787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002748918229797748,
+      "loss": 0.0116,
+      "step": 4788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027488108522541146,
+      "loss": 0.043,
+      "step": 4789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002748703453852983,
+      "loss": 0.0315,
+      "step": 4790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002748596034596146,
+      "loss": 0.0191,
+      "step": 4791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027484885944853985,
+      "loss": 0.0346,
+      "step": 4792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027483811335225354,
+      "loss": 0.0156,
+      "step": 4793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027482736517093505,
+      "loss": 0.0252,
+      "step": 4794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002748166149047639,
+      "loss": 0.0379,
+      "step": 4795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027480586255391976,
+      "loss": 0.0156,
+      "step": 4796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027479510811858205,
+      "loss": 0.0069,
+      "step": 4797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747843515989305,
+      "loss": 0.0175,
+      "step": 4798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747735929951447,
+      "loss": 0.007,
+      "step": 4799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747628323074044,
+      "loss": 0.0361,
+      "step": 4800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027475206953588927,
+      "loss": 0.0165,
+      "step": 4801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747413046807791,
+      "loss": 0.0351,
+      "step": 4802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027473053774225366,
+      "loss": 0.0044,
+      "step": 4803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027471976872049284,
+      "loss": 0.033,
+      "step": 4804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747089976156764,
+      "loss": 0.0109,
+      "step": 4805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027469822442798437,
+      "loss": 0.0431,
+      "step": 4806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027468744915759655,
+      "loss": 0.0088,
+      "step": 4807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027467667180469294,
+      "loss": 0.0034,
+      "step": 4808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746658923694536,
+      "loss": 0.0229,
+      "step": 4809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027465511085205855,
+      "loss": 0.011,
+      "step": 4810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027464432725268783,
+      "loss": 0.045,
+      "step": 4811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746335415715216,
+      "loss": 0.0135,
+      "step": 4812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027462275380873996,
+      "loss": 0.0116,
+      "step": 4813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000274611963964523,
+      "loss": 0.0108,
+      "step": 4814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746011720390511,
+      "loss": 0.0177,
+      "step": 4815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745903780325044,
+      "loss": 0.0167,
+      "step": 4816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745795819450632,
+      "loss": 0.0305,
+      "step": 4817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027456878377690784,
+      "loss": 0.0224,
+      "step": 4818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027455798352821865,
+      "loss": 0.0139,
+      "step": 4819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027454718119917604,
+      "loss": 0.002,
+      "step": 4820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745363767899603,
+      "loss": 0.0163,
+      "step": 4821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745255703007521,
+      "loss": 0.026,
+      "step": 4822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745147617317318,
+      "loss": 0.014,
+      "step": 4823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745039510830799,
+      "loss": 0.0083,
+      "step": 4824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000274493138354977,
+      "loss": 0.0137,
+      "step": 4825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744823235476037,
+      "loss": 0.0131,
+      "step": 4826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027447150666114065,
+      "loss": 0.0245,
+      "step": 4827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027446068769576847,
+      "loss": 0.0426,
+      "step": 4828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027444986665166787,
+      "loss": 0.0533,
+      "step": 4829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744390435290196,
+      "loss": 0.0103,
+      "step": 4830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027442821832800434,
+      "loss": 0.0081,
+      "step": 4831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000274417391048803,
+      "loss": 0.0127,
+      "step": 4832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744065616915964,
+      "loss": 0.0286,
+      "step": 4833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002743957302565653,
+      "loss": 0.0073,
+      "step": 4834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027438489674389075,
+      "loss": 0.0175,
+      "step": 4835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027437406115375364,
+      "loss": 0.0138,
+      "step": 4836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002743632234863349,
+      "loss": 0.0405,
+      "step": 4837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027435238374181567,
+      "loss": 0.0183,
+      "step": 4838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027434154192037684,
+      "loss": 0.0307,
+      "step": 4839
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002743306980221995,
+      "loss": 0.0166,
+      "step": 4840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027431985204746484,
+      "loss": 0.0174,
+      "step": 4841
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000274309003996354,
+      "loss": 0.0219,
+      "step": 4842
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027429815386904817,
+      "loss": 0.0225,
+      "step": 4843
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027428730166572854,
+      "loss": 0.0306,
+      "step": 4844
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002742764473865763,
+      "loss": 0.0186,
+      "step": 4845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027426559103177283,
+      "loss": 0.0811,
+      "step": 4846
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027425473260149946,
+      "loss": 0.0292,
+      "step": 4847
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002742438720959374,
+      "loss": 0.0364,
+      "step": 4848
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027423300951526827,
+      "loss": 0.0687,
+      "step": 4849
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002742221448596733,
+      "loss": 0.0387,
+      "step": 4850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002742112781293341,
+      "loss": 0.0108,
+      "step": 4851
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000274200409324432,
+      "loss": 0.0163,
+      "step": 4852
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741895384451487,
+      "loss": 0.0189,
+      "step": 4853
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741786654916656,
+      "loss": 0.0121,
+      "step": 4854
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027416779046416447,
+      "loss": 0.0259,
+      "step": 4855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741569133628268,
+      "loss": 0.0142,
+      "step": 4856
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027414603418783427,
+      "loss": 0.0222,
+      "step": 4857
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741351529393687,
+      "loss": 0.0146,
+      "step": 4858
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027412426961761176,
+      "loss": 0.0283,
+      "step": 4859
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741133842227452,
+      "loss": 0.0353,
+      "step": 4860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027410249675495084,
+      "loss": 0.0182,
+      "step": 4861
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027409160721441055,
+      "loss": 0.0212,
+      "step": 4862
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027408071560130614,
+      "loss": 0.013,
+      "step": 4863
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002740698219158196,
+      "loss": 0.0611,
+      "step": 4864
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027405892615813275,
+      "loss": 0.0162,
+      "step": 4865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002740480283284277,
+      "loss": 0.0059,
+      "step": 4866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027403712842688645,
+      "loss": 0.0155,
+      "step": 4867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000274026226453691,
+      "loss": 0.071,
+      "step": 4868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002740153224090234,
+      "loss": 0.0263,
+      "step": 4869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027400441629306594,
+      "loss": 0.0043,
+      "step": 4870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027399350810600057,
+      "loss": 0.0208,
+      "step": 4871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739825978480096,
+      "loss": 0.0133,
+      "step": 4872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739716855192751,
+      "loss": 0.017,
+      "step": 4873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027396077111997957,
+      "loss": 0.0446,
+      "step": 4874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739498546503051,
+      "loss": 0.0166,
+      "step": 4875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739389361104341,
+      "loss": 0.0137,
+      "step": 4876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000273928015500549,
+      "loss": 0.0234,
+      "step": 4877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000273917092820832,
+      "loss": 0.0125,
+      "step": 4878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739061680714657,
+      "loss": 0.0233,
+      "step": 4879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738952412526325,
+      "loss": 0.0032,
+      "step": 4880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027388431236451485,
+      "loss": 0.0195,
+      "step": 4881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738733814072954,
+      "loss": 0.0153,
+      "step": 4882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027386244838115663,
+      "loss": 0.0609,
+      "step": 4883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027385151328628113,
+      "loss": 0.0041,
+      "step": 4884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738405761228516,
+      "loss": 0.044,
+      "step": 4885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738296368910507,
+      "loss": 0.0258,
+      "step": 4886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002738186955910611,
+      "loss": 0.0395,
+      "step": 4887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027380775222306556,
+      "loss": 0.0248,
+      "step": 4888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027379680678724685,
+      "loss": 0.0501,
+      "step": 4889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027378585928378776,
+      "loss": 0.0148,
+      "step": 4890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737749097128712,
+      "loss": 0.0058,
+      "step": 4891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027376395807468003,
+      "loss": 0.0175,
+      "step": 4892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027375300436939705,
+      "loss": 0.0735,
+      "step": 4893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027374204859720536,
+      "loss": 0.0253,
+      "step": 4894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737310907582879,
+      "loss": 0.0172,
+      "step": 4895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027372013085282757,
+      "loss": 0.0254,
+      "step": 4896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737091688810076,
+      "loss": 0.0126,
+      "step": 4897
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273698204843011,
+      "loss": 0.0161,
+      "step": 4898
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027368723873902085,
+      "loss": 0.0149,
+      "step": 4899
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027367627056922034,
+      "loss": 0.0184,
+      "step": 4900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027366530033379264,
+      "loss": 0.0145,
+      "step": 4901
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273654328032921,
+      "loss": 0.0158,
+      "step": 4902
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002736433536667887,
+      "loss": 0.0131,
+      "step": 4903
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027363237723557893,
+      "loss": 0.0277,
+      "step": 4904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027362139873947513,
+      "loss": 0.0404,
+      "step": 4905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002736104181786607,
+      "loss": 0.0064,
+      "step": 4906
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735994355533188,
+      "loss": 0.01,
+      "step": 4907
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735884508636331,
+      "loss": 0.0311,
+      "step": 4908
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273577464109787,
+      "loss": 0.0178,
+      "step": 4909
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735664752919639,
+      "loss": 0.0294,
+      "step": 4910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735554844103475,
+      "loss": 0.0235,
+      "step": 4911
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735444914651212,
+      "loss": 0.0744,
+      "step": 4912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735334964564688,
+      "loss": 0.0094,
+      "step": 4913
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027352249938457373,
+      "loss": 0.0288,
+      "step": 4914
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735115002496198,
+      "loss": 0.0135,
+      "step": 4915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027350049905179065,
+      "loss": 0.0195,
+      "step": 4916
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027348949579127004,
+      "loss": 0.0156,
+      "step": 4917
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027347849046824177,
+      "loss": 0.0034,
+      "step": 4918
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002734674830828896,
+      "loss": 0.007,
+      "step": 4919
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002734564736353974,
+      "loss": 0.0282,
+      "step": 4920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002734454621259491,
+      "loss": 0.0118,
+      "step": 4921
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002734344485547285,
+      "loss": 0.0074,
+      "step": 4922
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027342343292191963,
+      "loss": 0.0154,
+      "step": 4923
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002734124152277065,
+      "loss": 0.004,
+      "step": 4924
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273401395472273,
+      "loss": 0.0045,
+      "step": 4925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733903736558033,
+      "loss": 0.0316,
+      "step": 4926
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027337934977848143,
+      "loss": 0.017,
+      "step": 4927
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733683238404915,
+      "loss": 0.0163,
+      "step": 4928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027335729584201777,
+      "loss": 0.0203,
+      "step": 4929
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733462657832443,
+      "loss": 0.0132,
+      "step": 4930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027333523366435537,
+      "loss": 0.0218,
+      "step": 4931
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027332419948553516,
+      "loss": 0.0072,
+      "step": 4932
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733131632469681,
+      "loss": 0.0136,
+      "step": 4933
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733021249488384,
+      "loss": 0.0069,
+      "step": 4934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002732910845913305,
+      "loss": 0.0321,
+      "step": 4935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027328004217462874,
+      "loss": 0.0134,
+      "step": 4936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002732689976989176,
+      "loss": 0.0253,
+      "step": 4937
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027325795116438147,
+      "loss": 0.0124,
+      "step": 4938
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002732469025712049,
+      "loss": 0.0075,
+      "step": 4939
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002732358519195724,
+      "loss": 0.0134,
+      "step": 4940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002732247992096686,
+      "loss": 0.0149,
+      "step": 4941
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027321374444167797,
+      "loss": 0.002,
+      "step": 4942
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027320268761578533,
+      "loss": 0.0109,
+      "step": 4943
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027319162873217515,
+      "loss": 0.0116,
+      "step": 4944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731805677910323,
+      "loss": 0.0128,
+      "step": 4945
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027316950479254134,
+      "loss": 0.0133,
+      "step": 4946
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731584397368873,
+      "loss": 0.0084,
+      "step": 4947
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027314737262425475,
+      "loss": 0.0323,
+      "step": 4948
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731363034548286,
+      "loss": 0.0155,
+      "step": 4949
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027312523222879374,
+      "loss": 0.0167,
+      "step": 4950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731141589463352,
+      "loss": 0.0101,
+      "step": 4951
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731030836076377,
+      "loss": 0.0072,
+      "step": 4952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730920062128864,
+      "loss": 0.0202,
+      "step": 4953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730809267622662,
+      "loss": 0.0232,
+      "step": 4954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730698452559622,
+      "loss": 0.012,
+      "step": 4955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730587616941595,
+      "loss": 0.0048,
+      "step": 4956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027304767607704315,
+      "loss": 0.0121,
+      "step": 4957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027303658840479835,
+      "loss": 0.0207,
+      "step": 4958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027302549867761026,
+      "loss": 0.0048,
+      "step": 4959
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027301440689566416,
+      "loss": 0.0032,
+      "step": 4960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027300331305914525,
+      "loss": 0.0071,
+      "step": 4961
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027299221716823875,
+      "loss": 0.0178,
+      "step": 4962
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729811192231301,
+      "loss": 0.0082,
+      "step": 4963
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729700192240047,
+      "loss": 0.0021,
+      "step": 4964
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729589171710478,
+      "loss": 0.0225,
+      "step": 4965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729478130644448,
+      "loss": 0.0101,
+      "step": 4966
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729367069043814,
+      "loss": 0.0282,
+      "step": 4967
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027292559869104285,
+      "loss": 0.0135,
+      "step": 4968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027291448842461474,
+      "loss": 0.0074,
+      "step": 4969
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729033761052827,
+      "loss": 0.0232,
+      "step": 4970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027289226173323226,
+      "loss": 0.0145,
+      "step": 4971
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027288114530864906,
+      "loss": 0.0062,
+      "step": 4972
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728700268317188,
+      "loss": 0.005,
+      "step": 4973
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728589063026272,
+      "loss": 0.0065,
+      "step": 4974
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027284778372155987,
+      "loss": 0.0035,
+      "step": 4975
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728366590887027,
+      "loss": 0.0114,
+      "step": 4976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027282553240424144,
+      "loss": 0.0062,
+      "step": 4977
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027281440366836194,
+      "loss": 0.0288,
+      "step": 4978
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027280327288125005,
+      "loss": 0.045,
+      "step": 4979
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027279214004309176,
+      "loss": 0.0053,
+      "step": 4980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727810051540729,
+      "loss": 0.0203,
+      "step": 4981
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727698682143795,
+      "loss": 0.092,
+      "step": 4982
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727587292241975,
+      "loss": 0.0201,
+      "step": 4983
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727475881837131,
+      "loss": 0.0184,
+      "step": 4984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002727364450931122,
+      "loss": 0.0095,
+      "step": 4985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027272529995258103,
+      "loss": 0.0128,
+      "step": 4986
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002727141527623057,
+      "loss": 0.0197,
+      "step": 4987
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002727030035224723,
+      "loss": 0.0303,
+      "step": 4988
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027269185223326723,
+      "loss": 0.0281,
+      "step": 4989
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027268069889487657,
+      "loss": 0.0525,
+      "step": 4990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027266954350748664,
+      "loss": 0.0241,
+      "step": 4991
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726583860712838,
+      "loss": 0.0035,
+      "step": 4992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027264722658645435,
+      "loss": 0.021,
+      "step": 4993
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726360650531848,
+      "loss": 0.0053,
+      "step": 4994
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726249014716614,
+      "loss": 0.0225,
+      "step": 4995
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726137358420707,
+      "loss": 0.0102,
+      "step": 4996
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726025681645991,
+      "loss": 0.0128,
+      "step": 4997
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027259139843943323,
+      "loss": 0.0241,
+      "step": 4998
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002725802266667596,
+      "loss": 0.0192,
+      "step": 4999
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002725690528467648,
+      "loss": 0.0061,
+      "step": 5000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027255787697963544,
+      "loss": 0.0431,
+      "step": 5001
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027254669906555824,
+      "loss": 0.0069,
+      "step": 5002
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027253551910471973,
+      "loss": 0.0077,
+      "step": 5003
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027252433709730684,
+      "loss": 0.0091,
+      "step": 5004
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027251315304350625,
+      "loss": 0.0097,
+      "step": 5005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002725019669435047,
+      "loss": 0.0092,
+      "step": 5006
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000272490778797489,
+      "loss": 0.0143,
+      "step": 5007
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002724795886056462,
+      "loss": 0.0142,
+      "step": 5008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000272468396368163,
+      "loss": 0.0059,
+      "step": 5009
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027245720208522647,
+      "loss": 0.0076,
+      "step": 5010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027244600575702347,
+      "loss": 0.0144,
+      "step": 5011
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027243480738374103,
+      "loss": 0.0189,
+      "step": 5012
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002724236069655662,
+      "loss": 0.0062,
+      "step": 5013
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027241240450268605,
+      "loss": 0.0203,
+      "step": 5014
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002724011999952877,
+      "loss": 0.005,
+      "step": 5015
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723899934435582,
+      "loss": 0.0122,
+      "step": 5016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723787848476848,
+      "loss": 0.0122,
+      "step": 5017
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027236757420785474,
+      "loss": 0.0192,
+      "step": 5018
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723563615242551,
+      "loss": 0.0241,
+      "step": 5019
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027234514679707337,
+      "loss": 0.0138,
+      "step": 5020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723339300264967,
+      "loss": 0.0264,
+      "step": 5021
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723227112127125,
+      "loss": 0.0067,
+      "step": 5022
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723114903559081,
+      "loss": 0.0179,
+      "step": 5023
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027230026745627095,
+      "loss": 0.024,
+      "step": 5024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722890425139884,
+      "loss": 0.0152,
+      "step": 5025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722778155292481,
+      "loss": 0.0078,
+      "step": 5026
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722665865022374,
+      "loss": 0.0072,
+      "step": 5027
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027225535543314396,
+      "loss": 0.0221,
+      "step": 5028
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027224412232215527,
+      "loss": 0.0192,
+      "step": 5029
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000272232887169459,
+      "loss": 0.0197,
+      "step": 5030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722216499752428,
+      "loss": 0.0198,
+      "step": 5031
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722104107396943,
+      "loss": 0.0014,
+      "step": 5032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027219916946300134,
+      "loss": 0.0142,
+      "step": 5033
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721879261453515,
+      "loss": 0.014,
+      "step": 5034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027217668078693264,
+      "loss": 0.0067,
+      "step": 5035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027216543338793263,
+      "loss": 0.0108,
+      "step": 5036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721541839485392,
+      "loss": 0.0047,
+      "step": 5037
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721429324689404,
+      "loss": 0.0033,
+      "step": 5038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027213167894932406,
+      "loss": 0.0179,
+      "step": 5039
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027212042338987816,
+      "loss": 0.0093,
+      "step": 5040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721091657907906,
+      "loss": 0.0171,
+      "step": 5041
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027209790615224954,
+      "loss": 0.0195,
+      "step": 5042
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720866444744429,
+      "loss": 0.0176,
+      "step": 5043
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720753807575589,
+      "loss": 0.0129,
+      "step": 5044
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720641150017856,
+      "loss": 0.0268,
+      "step": 5045
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720528472073111,
+      "loss": 0.0141,
+      "step": 5046
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027204157737432375,
+      "loss": 0.02,
+      "step": 5047
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720303055030117,
+      "loss": 0.0091,
+      "step": 5048
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027201903159356313,
+      "loss": 0.0187,
+      "step": 5049
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720077556461664,
+      "loss": 0.0145,
+      "step": 5050
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027199647766100986,
+      "loss": 0.0048,
+      "step": 5051
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719851976382819,
+      "loss": 0.0298,
+      "step": 5052
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027197391557817085,
+      "loss": 0.0078,
+      "step": 5053
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027196263148086515,
+      "loss": 0.0316,
+      "step": 5054
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719513453465533,
+      "loss": 0.0048,
+      "step": 5055
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719400571754238,
+      "loss": 0.0086,
+      "step": 5056
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027192876696766516,
+      "loss": 0.0537,
+      "step": 5057
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027191747472346593,
+      "loss": 0.0418,
+      "step": 5058
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719061804430148,
+      "loss": 0.0088,
+      "step": 5059
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027189488412650026,
+      "loss": 0.0047,
+      "step": 5060
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027188358577411113,
+      "loss": 0.005,
+      "step": 5061
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027187228538603597,
+      "loss": 0.0086,
+      "step": 5062
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002718609829624636,
+      "loss": 0.0099,
+      "step": 5063
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027184967850358286,
+      "loss": 0.0053,
+      "step": 5064
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002718383720095825,
+      "loss": 0.0109,
+      "step": 5065
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002718270634806512,
+      "loss": 0.0088,
+      "step": 5066
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027181575291697807,
+      "loss": 0.0064,
+      "step": 5067
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027180444031875195,
+      "loss": 0.0111,
+      "step": 5068
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717931256861617,
+      "loss": 0.0301,
+      "step": 5069
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027178180901939636,
+      "loss": 0.0031,
+      "step": 5070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027177049031864496,
+      "loss": 0.018,
+      "step": 5071
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717591695840965,
+      "loss": 0.0049,
+      "step": 5072
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027174784681594004,
+      "loss": 0.0137,
+      "step": 5073
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027173652201436475,
+      "loss": 0.0051,
+      "step": 5074
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027172519517955975,
+      "loss": 0.0531,
+      "step": 5075
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027171386631171424,
+      "loss": 0.0117,
+      "step": 5076
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027170253541101736,
+      "loss": 0.01,
+      "step": 5077
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716912024776584,
+      "loss": 0.0077,
+      "step": 5078
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716798675118267,
+      "loss": 0.0387,
+      "step": 5079
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716685305137115,
+      "loss": 0.0141,
+      "step": 5080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716571914835022,
+      "loss": 0.0431,
+      "step": 5081
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716458504213881,
+      "loss": 0.0436,
+      "step": 5082
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027163450732755874,
+      "loss": 0.02,
+      "step": 5083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002716231622022035,
+      "loss": 0.0449,
+      "step": 5084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002716118150455119,
+      "loss": 0.0128,
+      "step": 5085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027160046585767333,
+      "loss": 0.0112,
+      "step": 5086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027158911463887747,
+      "loss": 0.0321,
+      "step": 5087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002715777613893139,
+      "loss": 0.0152,
+      "step": 5088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002715664061091722,
+      "loss": 0.0046,
+      "step": 5089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027155504879864214,
+      "loss": 0.0154,
+      "step": 5090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027154368945791325,
+      "loss": 0.0137,
+      "step": 5091
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027153232808717527,
+      "loss": 0.0131,
+      "step": 5092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027152096468661805,
+      "loss": 0.0149,
+      "step": 5093
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027150959925643133,
+      "loss": 0.0047,
+      "step": 5094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714982317968049,
+      "loss": 0.0095,
+      "step": 5095
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714868623079287,
+      "loss": 0.0021,
+      "step": 5096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714754907899926,
+      "loss": 0.0069,
+      "step": 5097
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714641172431864,
+      "loss": 0.0093,
+      "step": 5098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714527416677003,
+      "loss": 0.0309,
+      "step": 5099
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714413640637241,
+      "loss": 0.0099,
+      "step": 5100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002714299844314478,
+      "loss": 0.0281,
+      "step": 5101
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027141860277106163,
+      "loss": 0.0026,
+      "step": 5102
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027140721908275556,
+      "loss": 0.0085,
+      "step": 5103
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027139583336671983,
+      "loss": 0.0123,
+      "step": 5104
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027138444562314445,
+      "loss": 0.0122,
+      "step": 5105
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027137305585221976,
+      "loss": 0.0123,
+      "step": 5106
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027136166405413586,
+      "loss": 0.0251,
+      "step": 5107
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027135027022908315,
+      "loss": 0.01,
+      "step": 5108
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027133887437725185,
+      "loss": 0.0254,
+      "step": 5109
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027132747649883233,
+      "loss": 0.029,
+      "step": 5110
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027131607659401484,
+      "loss": 0.0177,
+      "step": 5111
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027130467466298996,
+      "loss": 0.0304,
+      "step": 5112
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000271293270705948,
+      "loss": 0.0052,
+      "step": 5113
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027128186472307944,
+      "loss": 0.0131,
+      "step": 5114
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027127045671457485,
+      "loss": 0.0071,
+      "step": 5115
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002712590466806247,
+      "loss": 0.0133,
+      "step": 5116
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027124763462141956,
+      "loss": 0.0104,
+      "step": 5117
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027123622053715,
+      "loss": 0.0132,
+      "step": 5118
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002712248044280068,
+      "loss": 0.0115,
+      "step": 5119
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027121338629418053,
+      "loss": 0.0077,
+      "step": 5120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027120196613586185,
+      "loss": 0.0178,
+      "step": 5121
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711905439532416,
+      "loss": 0.0126,
+      "step": 5122
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711791197465105,
+      "loss": 0.0153,
+      "step": 5123
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711676935158593,
+      "loss": 0.0214,
+      "step": 5124
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027115626526147895,
+      "loss": 0.0214,
+      "step": 5125
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711448349835603,
+      "loss": 0.0199,
+      "step": 5126
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027113340268229415,
+      "loss": 0.0236,
+      "step": 5127
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027112196835787157,
+      "loss": 0.0353,
+      "step": 5128
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027111053201048344,
+      "loss": 0.0095,
+      "step": 5129
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002710990936403208,
+      "loss": 0.0053,
+      "step": 5130
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027108765324757484,
+      "loss": 0.0119,
+      "step": 5131
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002710762108324364,
+      "loss": 0.0143,
+      "step": 5132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710647663950967,
+      "loss": 0.018,
+      "step": 5133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710533199357469,
+      "loss": 0.0203,
+      "step": 5134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710418714545781,
+      "loss": 0.0119,
+      "step": 5135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710304209517816,
+      "loss": 0.0062,
+      "step": 5136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027101896842754864,
+      "loss": 0.0257,
+      "step": 5137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710075138820704,
+      "loss": 0.0056,
+      "step": 5138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709960573155383,
+      "loss": 0.026,
+      "step": 5139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027098459872814366,
+      "loss": 0.0095,
+      "step": 5140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709731381200778,
+      "loss": 0.0068,
+      "step": 5141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027096167549153224,
+      "loss": 0.0299,
+      "step": 5142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709502108426983,
+      "loss": 0.0082,
+      "step": 5143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027093874417376756,
+      "loss": 0.0508,
+      "step": 5144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027092727548493154,
+      "loss": 0.0086,
+      "step": 5145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709158047763817,
+      "loss": 0.0118,
+      "step": 5146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709043320483097,
+      "loss": 0.0351,
+      "step": 5147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027089285730090714,
+      "loss": 0.0186,
+      "step": 5148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708813805343656,
+      "loss": 0.04,
+      "step": 5149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027086990174887693,
+      "loss": 0.0245,
+      "step": 5150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708584209446327,
+      "loss": 0.0194,
+      "step": 5151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708469381218247,
+      "loss": 0.0071,
+      "step": 5152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027083545328064466,
+      "loss": 0.0108,
+      "step": 5153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027082396642128453,
+      "loss": 0.0079,
+      "step": 5154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708124775439361,
+      "loss": 0.0122,
+      "step": 5155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708009866487912,
+      "loss": 0.008,
+      "step": 5156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027078949373604176,
+      "loss": 0.0119,
+      "step": 5157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027077799880587984,
+      "loss": 0.0159,
+      "step": 5158
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027076650185849733,
+      "loss": 0.0138,
+      "step": 5159
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027075500289408625,
+      "loss": 0.0068,
+      "step": 5160
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707435019128387,
+      "loss": 0.0112,
+      "step": 5161
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027073199891494677,
+      "loss": 0.0039,
+      "step": 5162
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027072049390060256,
+      "loss": 0.0311,
+      "step": 5163
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707089868699982,
+      "loss": 0.007,
+      "step": 5164
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027069747782332596,
+      "loss": 0.0113,
+      "step": 5165
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027068596676077793,
+      "loss": 0.0114,
+      "step": 5166
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002706744536825465,
+      "loss": 0.0058,
+      "step": 5167
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027066293858882387,
+      "loss": 0.0104,
+      "step": 5168
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002706514214798024,
+      "loss": 0.0101,
+      "step": 5169
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027063990235567443,
+      "loss": 0.0144,
+      "step": 5170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027062838121663245,
+      "loss": 0.0088,
+      "step": 5171
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027061685806286875,
+      "loss": 0.0104,
+      "step": 5172
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002706053328945758,
+      "loss": 0.0102,
+      "step": 5173
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002705938057119462,
+      "loss": 0.0069,
+      "step": 5174
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027058227651517236,
+      "loss": 0.0183,
+      "step": 5175
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027057074530444694,
+      "loss": 0.0113,
+      "step": 5176
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002705592120799624,
+      "loss": 0.0157,
+      "step": 5177
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002705476768419115,
+      "loss": 0.0273,
+      "step": 5178
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027053613959048683,
+      "loss": 0.0322,
+      "step": 5179
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027052460032588114,
+      "loss": 0.008,
+      "step": 5180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027051305904828704,
+      "loss": 0.0114,
+      "step": 5181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002705015157578974,
+      "loss": 0.0086,
+      "step": 5182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000270489970454905,
+      "loss": 0.011,
+      "step": 5183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027047842313950265,
+      "loss": 0.0012,
+      "step": 5184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704668738118832,
+      "loss": 0.005,
+      "step": 5185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704553224722396,
+      "loss": 0.0084,
+      "step": 5186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027044376912076473,
+      "loss": 0.0175,
+      "step": 5187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704322137576516,
+      "loss": 0.0273,
+      "step": 5188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027042065638309306,
+      "loss": 0.0056,
+      "step": 5189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704090969972823,
+      "loss": 0.002,
+      "step": 5190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703975356004123,
+      "loss": 0.0109,
+      "step": 5191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027038597219267624,
+      "loss": 0.0037,
+      "step": 5192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703744067742672,
+      "loss": 0.0174,
+      "step": 5193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703628393453783,
+      "loss": 0.0143,
+      "step": 5194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027035126990620286,
+      "loss": 0.0141,
+      "step": 5195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000270339698456934,
+      "loss": 0.0356,
+      "step": 5196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000270328124997765,
+      "loss": 0.026,
+      "step": 5197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027031654952888913,
+      "loss": 0.013,
+      "step": 5198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027030497205049984,
+      "loss": 0.0076,
+      "step": 5199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002702933925627904,
+      "loss": 0.0084,
+      "step": 5200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027028181106595424,
+      "loss": 0.0124,
+      "step": 5201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027027022756018475,
+      "loss": 0.0077,
+      "step": 5202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027025864204567547,
+      "loss": 0.0208,
+      "step": 5203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027024705452261986,
+      "loss": 0.0077,
+      "step": 5204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002702354649912115,
+      "loss": 0.0315,
+      "step": 5205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027022387345164385,
+      "loss": 0.0134,
+      "step": 5206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002702122799041106,
+      "loss": 0.0253,
+      "step": 5207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027020068434880533,
+      "loss": 0.0168,
+      "step": 5208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027018908678592176,
+      "loss": 0.0209,
+      "step": 5209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701774872156536,
+      "loss": 0.0058,
+      "step": 5210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027016588563819456,
+      "loss": 0.0132,
+      "step": 5211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701542820537383,
+      "loss": 0.0228,
+      "step": 5212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701426764624788,
+      "loss": 0.0094,
+      "step": 5213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701310688646098,
+      "loss": 0.0185,
+      "step": 5214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701194592603252,
+      "loss": 0.0148,
+      "step": 5215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701078476498189,
+      "loss": 0.0029,
+      "step": 5216
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700962340332848,
+      "loss": 0.0178,
+      "step": 5217
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700846184109169,
+      "loss": 0.0257,
+      "step": 5218
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700730007829092,
+      "loss": 0.0115,
+      "step": 5219
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027006138114945574,
+      "loss": 0.0101,
+      "step": 5220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027004975951075053,
+      "loss": 0.0231,
+      "step": 5221
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027003813586698775,
+      "loss": 0.0315,
+      "step": 5222
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027002651021836157,
+      "loss": 0.0137,
+      "step": 5223
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000270014882565066,
+      "loss": 0.0323,
+      "step": 5224
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027000325290729547,
+      "loss": 0.0701,
+      "step": 5225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000269991621245244,
+      "loss": 0.0199,
+      "step": 5226
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000269979987579106,
+      "loss": 0.029,
+      "step": 5227
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002699683519090757,
+      "loss": 0.0279,
+      "step": 5228
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00026995671423534753,
+      "loss": 0.02,
+      "step": 5229
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00026994507455811576,
+      "loss": 0.0084,
+      "step": 5230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002699334328775748,
+      "loss": 0.0169,
+      "step": 5231
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026992178919391915,
+      "loss": 0.0177,
+      "step": 5232
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002699101435073432,
+      "loss": 0.0086,
+      "step": 5233
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698984958180416,
+      "loss": 0.0227,
+      "step": 5234
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026988684612620875,
+      "loss": 0.0162,
+      "step": 5235
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026987519443203925,
+      "loss": 0.0102,
+      "step": 5236
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026986354073572775,
+      "loss": 0.0144,
+      "step": 5237
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026985188503746886,
+      "loss": 0.0052,
+      "step": 5238
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698402273374572,
+      "loss": 0.0197,
+      "step": 5239
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698285676358876,
+      "loss": 0.0274,
+      "step": 5240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698169059329547,
+      "loss": 0.0303,
+      "step": 5241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698052422288533,
+      "loss": 0.0145,
+      "step": 5242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026979357652377826,
+      "loss": 0.0063,
+      "step": 5243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002697819088179243,
+      "loss": 0.0122,
+      "step": 5244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002697702391114864,
+      "loss": 0.0131,
+      "step": 5245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002697585674046594,
+      "loss": 0.0116,
+      "step": 5246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002697468936976382,
+      "loss": 0.0342,
+      "step": 5247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002697352179906179,
+      "loss": 0.0167,
+      "step": 5248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026972354028379343,
+      "loss": 0.0047,
+      "step": 5249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026971186057735987,
+      "loss": 0.0391,
+      "step": 5250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026970017887151227,
+      "loss": 0.0226,
+      "step": 5251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696884951664457,
+      "loss": 0.0082,
+      "step": 5252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696768094623554,
+      "loss": 0.007,
+      "step": 5253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026966512175943634,
+      "loss": 0.0035,
+      "step": 5254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026965343205788395,
+      "loss": 0.0101,
+      "step": 5255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026964174035789336,
+      "loss": 0.0398,
+      "step": 5256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026963004665965995,
+      "loss": 0.0352,
+      "step": 5257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026961835096337886,
+      "loss": 0.0114,
+      "step": 5258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026960665326924554,
+      "loss": 0.012,
+      "step": 5259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026959495357745534,
+      "loss": 0.0155,
+      "step": 5260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026958325188820365,
+      "loss": 0.007,
+      "step": 5261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695715482016859,
+      "loss": 0.0187,
+      "step": 5262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695598425180976,
+      "loss": 0.0015,
+      "step": 5263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026954813483763426,
+      "loss": 0.0169,
+      "step": 5264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695364251604914,
+      "loss": 0.0175,
+      "step": 5265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026952471348686463,
+      "loss": 0.0314,
+      "step": 5266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026951299981694946,
+      "loss": 0.007,
+      "step": 5267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026950128415094166,
+      "loss": 0.0156,
+      "step": 5268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694895664890368,
+      "loss": 0.0158,
+      "step": 5269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694778468314306,
+      "loss": 0.0133,
+      "step": 5270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026946612517831893,
+      "loss": 0.0099,
+      "step": 5271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026945440152989736,
+      "loss": 0.0253,
+      "step": 5272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026944267588636185,
+      "loss": 0.0085,
+      "step": 5273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026943094824790823,
+      "loss": 0.0126,
+      "step": 5274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026941921861473224,
+      "loss": 0.0097,
+      "step": 5275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694074869870299,
+      "loss": 0.02,
+      "step": 5276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002693957533649972,
+      "loss": 0.008,
+      "step": 5277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026938401774883,
+      "loss": 0.0217,
+      "step": 5278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026937228013872427,
+      "loss": 0.0079,
+      "step": 5279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693605405348762,
+      "loss": 0.0161,
+      "step": 5280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026934879893748186,
+      "loss": 0.0298,
+      "step": 5281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693370553467372,
+      "loss": 0.0219,
+      "step": 5282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693253097628385,
+      "loss": 0.0082,
+      "step": 5283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693135621859818,
+      "loss": 0.0074,
+      "step": 5284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693018126163635,
+      "loss": 0.0362,
+      "step": 5285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026929006105417966,
+      "loss": 0.0302,
+      "step": 5286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026927830749962664,
+      "loss": 0.0091,
+      "step": 5287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692665519529008,
+      "loss": 0.0186,
+      "step": 5288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692547944141983,
+      "loss": 0.0067,
+      "step": 5289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692430348837157,
+      "loss": 0.0047,
+      "step": 5290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692312733616493,
+      "loss": 0.0034,
+      "step": 5291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692195098481956,
+      "loss": 0.0077,
+      "step": 5292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000269207744343551,
+      "loss": 0.0125,
+      "step": 5293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026919597684791204,
+      "loss": 0.0091,
+      "step": 5294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691842073614753,
+      "loss": 0.0443,
+      "step": 5295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026917243588443735,
+      "loss": 0.022,
+      "step": 5296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691606624169947,
+      "loss": 0.0094,
+      "step": 5297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691488869593441,
+      "loss": 0.0215,
+      "step": 5298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026913710951168213,
+      "loss": 0.0116,
+      "step": 5299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691253300742056,
+      "loss": 0.0176,
+      "step": 5300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691135486471112,
+      "loss": 0.0164,
+      "step": 5301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691017652305957,
+      "loss": 0.0175,
+      "step": 5302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690899798248558,
+      "loss": 0.0017,
+      "step": 5303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026907819243008853,
+      "loss": 0.0127,
+      "step": 5304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026906640304649065,
+      "loss": 0.0153,
+      "step": 5305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690546116742591,
+      "loss": 0.009,
+      "step": 5306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026904281831359076,
+      "loss": 0.0064,
+      "step": 5307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690310229646827,
+      "loss": 0.0126,
+      "step": 5308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026901922562773184,
+      "loss": 0.0085,
+      "step": 5309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690074263029352,
+      "loss": 0.0389,
+      "step": 5310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026899562499049,
+      "loss": 0.0327,
+      "step": 5311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002689838216905932,
+      "loss": 0.0081,
+      "step": 5312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026897201640344194,
+      "loss": 0.0184,
+      "step": 5313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026896020912923346,
+      "loss": 0.0247,
+      "step": 5314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002689483998681649,
+      "loss": 0.0065,
+      "step": 5315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002689365886204336,
+      "loss": 0.0251,
+      "step": 5316
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002689247753862367,
+      "loss": 0.0314,
+      "step": 5317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026891296016577166,
+      "loss": 0.0123,
+      "step": 5318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002689011429592356,
+      "loss": 0.0111,
+      "step": 5319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688893237668261,
+      "loss": 0.0047,
+      "step": 5320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688775025887404,
+      "loss": 0.0079,
+      "step": 5321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688656794251761,
+      "loss": 0.0127,
+      "step": 5322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688538542763305,
+      "loss": 0.0151,
+      "step": 5323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026884202714240116,
+      "loss": 0.0272,
+      "step": 5324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688301980235857,
+      "loss": 0.0217,
+      "step": 5325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688183669200816,
+      "loss": 0.0321,
+      "step": 5326
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688065338320865,
+      "loss": 0.021,
+      "step": 5327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000268794698759798,
+      "loss": 0.0216,
+      "step": 5328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002687828617034139,
+      "loss": 0.0285,
+      "step": 5329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026877102266313163,
+      "loss": 0.0139,
+      "step": 5330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002687591816391491,
+      "loss": 0.0136,
+      "step": 5331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026874733863166417,
+      "loss": 0.0271,
+      "step": 5332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026873549364087447,
+      "loss": 0.0046,
+      "step": 5333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002687236466669779,
+      "loss": 0.0123,
+      "step": 5334
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026871179771017237,
+      "loss": 0.0112,
+      "step": 5335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026869994677065565,
+      "loss": 0.0141,
+      "step": 5336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686880938486258,
+      "loss": 0.0075,
+      "step": 5337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686762389442808,
+      "loss": 0.0167,
+      "step": 5338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686643820578185,
+      "loss": 0.011,
+      "step": 5339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026865252318943706,
+      "loss": 0.0087,
+      "step": 5340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686406623393345,
+      "loss": 0.025,
+      "step": 5341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026862879950770893,
+      "loss": 0.0077,
+      "step": 5342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686169346947585,
+      "loss": 0.0216,
+      "step": 5343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026860506790068133,
+      "loss": 0.0157,
+      "step": 5344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026859319912567566,
+      "loss": 0.0061,
+      "step": 5345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026858132836993974,
+      "loss": 0.0212,
+      "step": 5346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002685694556336717,
+      "loss": 0.0219,
+      "step": 5347
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026855758091707003,
+      "loss": 0.0128,
+      "step": 5348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026854570422033295,
+      "loss": 0.0189,
+      "step": 5349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002685338255436588,
+      "loss": 0.0092,
+      "step": 5350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026852194488724606,
+      "loss": 0.0207,
+      "step": 5351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026851006225129306,
+      "loss": 0.0037,
+      "step": 5352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026849817763599835,
+      "loss": 0.0054,
+      "step": 5353
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684862910415604,
+      "loss": 0.0091,
+      "step": 5354
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026847440246817775,
+      "loss": 0.0154,
+      "step": 5355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684625119160489,
+      "loss": 0.0195,
+      "step": 5356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026845061938537247,
+      "loss": 0.0128,
+      "step": 5357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684387248763471,
+      "loss": 0.0042,
+      "step": 5358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026842682838917154,
+      "loss": 0.0087,
+      "step": 5359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684149299240443,
+      "loss": 0.0231,
+      "step": 5360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026840302948116427,
+      "loss": 0.0169,
+      "step": 5361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026839112706073014,
+      "loss": 0.0339,
+      "step": 5362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683792226629407,
+      "loss": 0.0112,
+      "step": 5363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683673162879948,
+      "loss": 0.007,
+      "step": 5364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683554079360913,
+      "loss": 0.0178,
+      "step": 5365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000268343497607429,
+      "loss": 0.0064,
+      "step": 5366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000268331585302207,
+      "loss": 0.0139,
+      "step": 5367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026831967102062404,
+      "loss": 0.0597,
+      "step": 5368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026830775476287933,
+      "loss": 0.0236,
+      "step": 5369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682958365291717,
+      "loss": 0.011,
+      "step": 5370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682839163197004,
+      "loss": 0.0376,
+      "step": 5371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682719941346644,
+      "loss": 0.0184,
+      "step": 5372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682600699742628,
+      "loss": 0.0173,
+      "step": 5373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682481438386948,
+      "loss": 0.0092,
+      "step": 5374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026823621572815966,
+      "loss": 0.0146,
+      "step": 5375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682242856428565,
+      "loss": 0.007,
+      "step": 5376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026821235358298455,
+      "loss": 0.0149,
+      "step": 5377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002682004195487432,
+      "loss": 0.0106,
+      "step": 5378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026818848354033173,
+      "loss": 0.0284,
+      "step": 5379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026817654555794947,
+      "loss": 0.0066,
+      "step": 5380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681646056017958,
+      "loss": 0.0072,
+      "step": 5381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681526636720702,
+      "loss": 0.0138,
+      "step": 5382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026814071976897206,
+      "loss": 0.0083,
+      "step": 5383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681287738927009,
+      "loss": 0.0122,
+      "step": 5384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681168260434563,
+      "loss": 0.0158,
+      "step": 5385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026810487622143766,
+      "loss": 0.013,
+      "step": 5386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680929244268447,
+      "loss": 0.0275,
+      "step": 5387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000268080970659877,
+      "loss": 0.0326,
+      "step": 5388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026806901492073415,
+      "loss": 0.0155,
+      "step": 5389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680570572096159,
+      "loss": 0.0086,
+      "step": 5390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680450975267219,
+      "loss": 0.0159,
+      "step": 5391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000268033135872252,
+      "loss": 0.0258,
+      "step": 5392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680211722464059,
+      "loss": 0.0238,
+      "step": 5393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680092066493834,
+      "loss": 0.0417,
+      "step": 5394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679972390813845,
+      "loss": 0.0495,
+      "step": 5395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679852695426089,
+      "loss": 0.0054,
+      "step": 5396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679732980332566,
+      "loss": 0.0077,
+      "step": 5397
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026796132455352754,
+      "loss": 0.0057,
+      "step": 5398
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026794934910362174,
+      "loss": 0.011,
+      "step": 5399
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026793737168373906,
+      "loss": 0.0147,
+      "step": 5400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679253922940797,
+      "loss": 0.0111,
+      "step": 5401
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679134109348437,
+      "loss": 0.0059,
+      "step": 5402
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026790142760623115,
+      "loss": 0.0102,
+      "step": 5403
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678894423084422,
+      "loss": 0.019,
+      "step": 5404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026787745504167705,
+      "loss": 0.0262,
+      "step": 5405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678654658061359,
+      "loss": 0.0095,
+      "step": 5406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026785347460201895,
+      "loss": 0.0147,
+      "step": 5407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678414814295265,
+      "loss": 0.0148,
+      "step": 5408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026782948628885887,
+      "loss": 0.0261,
+      "step": 5409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026781748918021646,
+      "loss": 0.0109,
+      "step": 5410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678054901037995,
+      "loss": 0.0045,
+      "step": 5411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026779348905980855,
+      "loss": 0.0261,
+      "step": 5412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000267781486048444,
+      "loss": 0.0287,
+      "step": 5413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026776948106990625,
+      "loss": 0.0033,
+      "step": 5414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026775747412439586,
+      "loss": 0.0142,
+      "step": 5415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026774546521211336,
+      "loss": 0.0113,
+      "step": 5416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026773345433325936,
+      "loss": 0.0218,
+      "step": 5417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026772144148803445,
+      "loss": 0.0198,
+      "step": 5418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002677094266766392,
+      "loss": 0.0064,
+      "step": 5419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026769740989927436,
+      "loss": 0.0034,
+      "step": 5420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002676853911561406,
+      "loss": 0.0182,
+      "step": 5421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002676733704474387,
+      "loss": 0.0141,
+      "step": 5422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026766134777336933,
+      "loss": 0.0041,
+      "step": 5423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002676493231341334,
+      "loss": 0.0216,
+      "step": 5424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026763729652993167,
+      "loss": 0.0166,
+      "step": 5425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000267625267960965,
+      "loss": 0.0033,
+      "step": 5426
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026761323742743436,
+      "loss": 0.0197,
+      "step": 5427
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026760120492954064,
+      "loss": 0.0324,
+      "step": 5428
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026758917046748484,
+      "loss": 0.0225,
+      "step": 5429
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675771340414679,
+      "loss": 0.0032,
+      "step": 5430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675650956516909,
+      "loss": 0.007,
+      "step": 5431
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026755305529835483,
+      "loss": 0.0232,
+      "step": 5432
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675410129816609,
+      "loss": 0.0054,
+      "step": 5433
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675289687018101,
+      "loss": 0.0129,
+      "step": 5434
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675169224590037,
+      "loss": 0.0189,
+      "step": 5435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675048742534429,
+      "loss": 0.0298,
+      "step": 5436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002674928240853289,
+      "loss": 0.023,
+      "step": 5437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026748077195486294,
+      "loss": 0.0122,
+      "step": 5438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026746871786224635,
+      "loss": 0.0082,
+      "step": 5439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026745666180768034,
+      "loss": 0.046,
+      "step": 5440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026744460379136644,
+      "loss": 0.0062,
+      "step": 5441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026743254381350596,
+      "loss": 0.0137,
+      "step": 5442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002674204818743003,
+      "loss": 0.0088,
+      "step": 5443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267408417973951,
+      "loss": 0.0284,
+      "step": 5444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026739635211265944,
+      "loss": 0.0098,
+      "step": 5445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026738428429062724,
+      "loss": 0.0054,
+      "step": 5446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673722145080559,
+      "loss": 0.0489,
+      "step": 5447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267360142765147,
+      "loss": 0.0155,
+      "step": 5448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673480690621022,
+      "loss": 0.0091,
+      "step": 5449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026733599339912316,
+      "loss": 0.0178,
+      "step": 5450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673239157764115,
+      "loss": 0.0216,
+      "step": 5451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267311836194169,
+      "loss": 0.0281,
+      "step": 5452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672997546525974,
+      "loss": 0.0048,
+      "step": 5453
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672876711518985,
+      "loss": 0.0043,
+      "step": 5454
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267275585692274,
+      "loss": 0.0126,
+      "step": 5455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672634982739259,
+      "loss": 0.001,
+      "step": 5456
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026725140889705606,
+      "loss": 0.0082,
+      "step": 5457
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672393175618664,
+      "loss": 0.0053,
+      "step": 5458
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672272242685588,
+      "loss": 0.0015,
+      "step": 5459
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672151290173352,
+      "loss": 0.0159,
+      "step": 5460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026720303180839775,
+      "loss": 0.0109,
+      "step": 5461
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671909326419484,
+      "loss": 0.0225,
+      "step": 5462
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671788315181893,
+      "loss": 0.017,
+      "step": 5463
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671667284373225,
+      "loss": 0.0025,
+      "step": 5464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026715462339955016,
+      "loss": 0.0141,
+      "step": 5465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026714251640507446,
+      "loss": 0.0194,
+      "step": 5466
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671304074540976,
+      "loss": 0.0126,
+      "step": 5467
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671182965468219,
+      "loss": 0.0124,
+      "step": 5468
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026710618368344955,
+      "loss": 0.0031,
+      "step": 5469
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026709406886418286,
+      "loss": 0.0166,
+      "step": 5470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670819520892242,
+      "loss": 0.0153,
+      "step": 5471
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670698333587759,
+      "loss": 0.0252,
+      "step": 5472
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026705771267304044,
+      "loss": 0.0267,
+      "step": 5473
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670455900322202,
+      "loss": 0.0228,
+      "step": 5474
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670334654365177,
+      "loss": 0.018,
+      "step": 5475
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026702133888613534,
+      "loss": 0.0268,
+      "step": 5476
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002670092103812758,
+      "loss": 0.0248,
+      "step": 5477
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669970799221415,
+      "loss": 0.0072,
+      "step": 5478
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026698494750893514,
+      "loss": 0.0093,
+      "step": 5479
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026697281314185943,
+      "loss": 0.0157,
+      "step": 5480
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026696067682111684,
+      "loss": 0.0154,
+      "step": 5481
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669485385469101,
+      "loss": 0.0102,
+      "step": 5482
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026693639831944216,
+      "loss": 0.0066,
+      "step": 5483
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669242561389156,
+      "loss": 0.0187,
+      "step": 5484
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669121120055332,
+      "loss": 0.0032,
+      "step": 5485
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668999659194979,
+      "loss": 0.008,
+      "step": 5486
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668878178810125,
+      "loss": 0.0172,
+      "step": 5487
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026687566789027983,
+      "loss": 0.0298,
+      "step": 5488
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668635159475029,
+      "loss": 0.0109,
+      "step": 5489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668513620528847,
+      "loss": 0.0283,
+      "step": 5490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026683920620662816,
+      "loss": 0.0186,
+      "step": 5491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026682704840893634,
+      "loss": 0.0262,
+      "step": 5492
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668148886600123,
+      "loss": 0.0093,
+      "step": 5493
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026680272696005906,
+      "loss": 0.011,
+      "step": 5494
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026679056330927985,
+      "loss": 0.0173,
+      "step": 5495
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026677839770787777,
+      "loss": 0.013,
+      "step": 5496
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000266766230156056,
+      "loss": 0.0073,
+      "step": 5497
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002667540606540178,
+      "loss": 0.0249,
+      "step": 5498
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026674188920196637,
+      "loss": 0.0255,
+      "step": 5499
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000266729715800105,
+      "loss": 0.0134,
+      "step": 5500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026671754044863704,
+      "loss": 0.0034,
+      "step": 5501
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026670536314776593,
+      "loss": 0.0305,
+      "step": 5502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666931838976949,
+      "loss": 0.0064,
+      "step": 5503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666810026986274,
+      "loss": 0.0068,
+      "step": 5504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666688195507669,
+      "loss": 0.0044,
+      "step": 5505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666566344543169,
+      "loss": 0.0077,
+      "step": 5506
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666444474094809,
+      "loss": 0.0093,
+      "step": 5507
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026663225841646243,
+      "loss": 0.0202,
+      "step": 5508
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666200674754651,
+      "loss": 0.0156,
+      "step": 5509
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666078745866925,
+      "loss": 0.0109,
+      "step": 5510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665956797503483,
+      "loss": 0.0112,
+      "step": 5511
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026658348296663617,
+      "loss": 0.0023,
+      "step": 5512
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026657128423575977,
+      "loss": 0.0308,
+      "step": 5513
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665590835579229,
+      "loss": 0.0057,
+      "step": 5514
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665468809333293,
+      "loss": 0.0541,
+      "step": 5515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665346763621828,
+      "loss": 0.0104,
+      "step": 5516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665224698446873,
+      "loss": 0.0097,
+      "step": 5517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665102613810465,
+      "loss": 0.0129,
+      "step": 5518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026649805097146446,
+      "loss": 0.0172,
+      "step": 5519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000266485838616145,
+      "loss": 0.0255,
+      "step": 5520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002664736243152922,
+      "loss": 0.0289,
+      "step": 5521
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026646140806911003,
+      "loss": 0.0136,
+      "step": 5522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002664491898778025,
+      "loss": 0.0134,
+      "step": 5523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002664369697415736,
+      "loss": 0.0105,
+      "step": 5524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026642474766062766,
+      "loss": 0.02,
+      "step": 5525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002664125236351686,
+      "loss": 0.0027,
+      "step": 5526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002664002976654006,
+      "loss": 0.0195,
+      "step": 5527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026638806975152796,
+      "loss": 0.0464,
+      "step": 5528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026637583989375487,
+      "loss": 0.0063,
+      "step": 5529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026636360809228555,
+      "loss": 0.0463,
+      "step": 5530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026635137434732436,
+      "loss": 0.0032,
+      "step": 5531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002663391386590755,
+      "loss": 0.0166,
+      "step": 5532
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026632690102774357,
+      "loss": 0.0133,
+      "step": 5533
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026631466145353265,
+      "loss": 0.0104,
+      "step": 5534
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002663024199366474,
+      "loss": 0.0088,
+      "step": 5535
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662901764772922,
+      "loss": 0.0182,
+      "step": 5536
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026627793107567157,
+      "loss": 0.0031,
+      "step": 5537
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026626568373198993,
+      "loss": 0.0135,
+      "step": 5538
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026625343444645194,
+      "loss": 0.0253,
+      "step": 5539
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662411832192622,
+      "loss": 0.0403,
+      "step": 5540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662289300506252,
+      "loss": 0.0291,
+      "step": 5541
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662166749407457,
+      "loss": 0.0089,
+      "step": 5542
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026620441788982835,
+      "loss": 0.0066,
+      "step": 5543
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026619215889807786,
+      "loss": 0.0052,
+      "step": 5544
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026617989796569905,
+      "loss": 0.0041,
+      "step": 5545
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026616763509289655,
+      "loss": 0.0236,
+      "step": 5546
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026615537027987535,
+      "loss": 0.0164,
+      "step": 5547
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026614310352684014,
+      "loss": 0.0136,
+      "step": 5548
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026613083483399593,
+      "loss": 0.0151,
+      "step": 5549
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002661185642015475,
+      "loss": 0.0163,
+      "step": 5550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002661062916296999,
+      "loss": 0.0035,
+      "step": 5551
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026609401711865803,
+      "loss": 0.0052,
+      "step": 5552
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026608174066862695,
+      "loss": 0.0041,
+      "step": 5553
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026606946227981167,
+      "loss": 0.0098,
+      "step": 5554
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002660571819524173,
+      "loss": 0.0108,
+      "step": 5555
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002660448996866489,
+      "loss": 0.0102,
+      "step": 5556
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002660326154827116,
+      "loss": 0.0463,
+      "step": 5557
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026602032934081065,
+      "loss": 0.0251,
+      "step": 5558
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026600804126115116,
+      "loss": 0.0104,
+      "step": 5559
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002659957512439384,
+      "loss": 0.0181,
+      "step": 5560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026598345928937765,
+      "loss": 0.0084,
+      "step": 5561
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026597116539767417,
+      "loss": 0.0023,
+      "step": 5562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002659588695690334,
+      "loss": 0.0216,
+      "step": 5563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002659465718036605,
+      "loss": 0.0323,
+      "step": 5564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000265934272101761,
+      "loss": 0.0384,
+      "step": 5565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002659219704635404,
+      "loss": 0.0045,
+      "step": 5566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000265909666889204,
+      "loss": 0.0018,
+      "step": 5567
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026589736137895743,
+      "loss": 0.0287,
+      "step": 5568
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658850539330061,
+      "loss": 0.0134,
+      "step": 5569
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658727445515556,
+      "loss": 0.0152,
+      "step": 5570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658604332348116,
+      "loss": 0.0047,
+      "step": 5571
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026584811998297967,
+      "loss": 0.0248,
+      "step": 5572
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026583580479626546,
+      "loss": 0.0132,
+      "step": 5573
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658234876748746,
+      "loss": 0.0014,
+      "step": 5574
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002658111686190129,
+      "loss": 0.0072,
+      "step": 5575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026579884762888606,
+      "loss": 0.0026,
+      "step": 5576
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657865247046999,
+      "loss": 0.0065,
+      "step": 5577
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657741998466602,
+      "loss": 0.0077,
+      "step": 5578
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657618730549728,
+      "loss": 0.0299,
+      "step": 5579
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026574954432984364,
+      "loss": 0.0166,
+      "step": 5580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026573721367147856,
+      "loss": 0.0094,
+      "step": 5581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657248810800836,
+      "loss": 0.014,
+      "step": 5582
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657125465558647,
+      "loss": 0.0073,
+      "step": 5583
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026570021009902776,
+      "loss": 0.0067,
+      "step": 5584
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002656878717097789,
+      "loss": 0.015,
+      "step": 5585
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002656755313883243,
+      "loss": 0.0416,
+      "step": 5586
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002656631891348699,
+      "loss": 0.0192,
+      "step": 5587
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026565084494962197,
+      "loss": 0.0063,
+      "step": 5588
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026563849883278655,
+      "loss": 0.0113,
+      "step": 5589
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026562615078457,
+      "loss": 0.0198,
+      "step": 5590
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002656138008051784,
+      "loss": 0.0147,
+      "step": 5591
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026560144889481815,
+      "loss": 0.0116,
+      "step": 5592
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026558909505369543,
+      "loss": 0.0155,
+      "step": 5593
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655767392820167,
+      "loss": 0.0306,
+      "step": 5594
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655643815799882,
+      "loss": 0.0152,
+      "step": 5595
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655520219478164,
+      "loss": 0.0128,
+      "step": 5596
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655396603857077,
+      "loss": 0.0158,
+      "step": 5597
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655272968938686,
+      "loss": 0.0125,
+      "step": 5598
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655149314725056,
+      "loss": 0.0188,
+      "step": 5599
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026550256412182513,
+      "loss": 0.0237,
+      "step": 5600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026549019484203385,
+      "loss": 0.0476,
+      "step": 5601
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654778236333383,
+      "loss": 0.0071,
+      "step": 5602
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654654504959451,
+      "loss": 0.0132,
+      "step": 5603
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026545307543006093,
+      "loss": 0.0278,
+      "step": 5604
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654406984358925,
+      "loss": 0.0123,
+      "step": 5605
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026542831951364645,
+      "loss": 0.0077,
+      "step": 5606
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654159386635296,
+      "loss": 0.0181,
+      "step": 5607
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654035558857487,
+      "loss": 0.0083,
+      "step": 5608
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026539117118051065,
+      "loss": 0.007,
+      "step": 5609
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653787845480221,
+      "loss": 0.0182,
+      "step": 5610
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653663959884901,
+      "loss": 0.0229,
+      "step": 5611
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653540055021216,
+      "loss": 0.0323,
+      "step": 5612
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026534161308912335,
+      "loss": 0.0165,
+      "step": 5613
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653292187497025,
+      "loss": 0.019,
+      "step": 5614
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026531682248406597,
+      "loss": 0.0289,
+      "step": 5615
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026530442429242087,
+      "loss": 0.0094,
+      "step": 5616
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652920241749742,
+      "loss": 0.0116,
+      "step": 5617
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652796221319331,
+      "loss": 0.0254,
+      "step": 5618
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652672181635047,
+      "loss": 0.044,
+      "step": 5619
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652548122698962,
+      "loss": 0.0061,
+      "step": 5620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026524240445131473,
+      "loss": 0.0104,
+      "step": 5621
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652299947079676,
+      "loss": 0.0125,
+      "step": 5622
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652175830400621,
+      "loss": 0.0071,
+      "step": 5623
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026520516944780536,
+      "loss": 0.0137,
+      "step": 5624
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026519275393140486,
+      "loss": 0.0216,
+      "step": 5625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000265180336491068,
+      "loss": 0.0229,
+      "step": 5626
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000265167917127002,
+      "loss": 0.0247,
+      "step": 5627
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026515549583941444,
+      "loss": 0.0109,
+      "step": 5628
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026514307262851276,
+      "loss": 0.0399,
+      "step": 5629
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002651306474945044,
+      "loss": 0.0313,
+      "step": 5630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002651182204375969,
+      "loss": 0.0159,
+      "step": 5631
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002651057914579978,
+      "loss": 0.0189,
+      "step": 5632
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650933605559147,
+      "loss": 0.0112,
+      "step": 5633
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026508092773155526,
+      "loss": 0.0082,
+      "step": 5634
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650684929851271,
+      "loss": 0.0145,
+      "step": 5635
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650560563168378,
+      "loss": 0.0049,
+      "step": 5636
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026504361772689526,
+      "loss": 0.0124,
+      "step": 5637
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650311772155071,
+      "loss": 0.0049,
+      "step": 5638
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026501873478288124,
+      "loss": 0.0178,
+      "step": 5639
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026500629042922535,
+      "loss": 0.0138,
+      "step": 5640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649938441547473,
+      "loss": 0.0303,
+      "step": 5641
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000264981395959655,
+      "loss": 0.0055,
+      "step": 5642
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649689458441563,
+      "loss": 0.0102,
+      "step": 5643
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649564938084592,
+      "loss": 0.0123,
+      "step": 5644
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026494403985277167,
+      "loss": 0.0195,
+      "step": 5645
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649315839773018,
+      "loss": 0.003,
+      "step": 5646
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026491912618225737,
+      "loss": 0.0137,
+      "step": 5647
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026490666646784665,
+      "loss": 0.0237,
+      "step": 5648
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026489420483427774,
+      "loss": 0.0239,
+      "step": 5649
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026488174128175875,
+      "loss": 0.0188,
+      "step": 5650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002648692758104978,
+      "loss": 0.0206,
+      "step": 5651
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026485680842070305,
+      "loss": 0.0236,
+      "step": 5652
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002648443391125829,
+      "loss": 0.0033,
+      "step": 5653
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002648318678863454,
+      "loss": 0.0109,
+      "step": 5654
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026481939474219907,
+      "loss": 0.0569,
+      "step": 5655
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000264806919680352,
+      "loss": 0.0191,
+      "step": 5656
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647944427010127,
+      "loss": 0.0128,
+      "step": 5657
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647819638043895,
+      "loss": 0.0238,
+      "step": 5658
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026476948299069083,
+      "loss": 0.0409,
+      "step": 5659
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026475700026012516,
+      "loss": 0.0135,
+      "step": 5660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000264744515612901,
+      "loss": 0.011,
+      "step": 5661
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026473202904922684,
+      "loss": 0.005,
+      "step": 5662
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026471954056931117,
+      "loss": 0.0288,
+      "step": 5663
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026470705017336263,
+      "loss": 0.0148,
+      "step": 5664
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002646945578615899,
+      "loss": 0.0222,
+      "step": 5665
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026468206363420145,
+      "loss": 0.0061,
+      "step": 5666
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002646695674914061,
+      "loss": 0.0066,
+      "step": 5667
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026465706943341254,
+      "loss": 0.0159,
+      "step": 5668
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026464456946042946,
+      "loss": 0.0078,
+      "step": 5669
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026463206757266567,
+      "loss": 0.0173,
+      "step": 5670
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026461956377033,
+      "loss": 0.019,
+      "step": 5671
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002646070580536312,
+      "loss": 0.0071,
+      "step": 5672
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026459455042277826,
+      "loss": 0.0123,
+      "step": 5673
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026458204087797995,
+      "loss": 0.0046,
+      "step": 5674
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645695294194453,
+      "loss": 0.0256,
+      "step": 5675
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026455701604738327,
+      "loss": 0.0202,
+      "step": 5676
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645445007620028,
+      "loss": 0.003,
+      "step": 5677
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026453198356351295,
+      "loss": 0.0062,
+      "step": 5678
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026451946445212276,
+      "loss": 0.0149,
+      "step": 5679
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026450694342804133,
+      "loss": 0.0021,
+      "step": 5680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002644944204914779,
+      "loss": 0.0067,
+      "step": 5681
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026448189564264144,
+      "loss": 0.0129,
+      "step": 5682
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026446936888174127,
+      "loss": 0.0142,
+      "step": 5683
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002644568402089865,
+      "loss": 0.0108,
+      "step": 5684
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002644443096245865,
+      "loss": 0.0295,
+      "step": 5685
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026443177712875043,
+      "loss": 0.0165,
+      "step": 5686
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026441924272168775,
+      "loss": 0.0188,
+      "step": 5687
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026440670640360777,
+      "loss": 0.025,
+      "step": 5688
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026439416817471975,
+      "loss": 0.0068,
+      "step": 5689
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002643816280352332,
+      "loss": 0.0051,
+      "step": 5690
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026436908598535754,
+      "loss": 0.0147,
+      "step": 5691
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026435654202530234,
+      "loss": 0.0043,
+      "step": 5692
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002643439961552769,
+      "loss": 0.0146,
+      "step": 5693
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264331448375491,
+      "loss": 0.0333,
+      "step": 5694
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264318898686154,
+      "loss": 0.0066,
+      "step": 5695
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026430634708747564,
+      "loss": 0.007,
+      "step": 5696
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642937935796656,
+      "loss": 0.0045,
+      "step": 5697
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642812381629334,
+      "loss": 0.0124,
+      "step": 5698
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642686808374887,
+      "loss": 0.0189,
+      "step": 5699
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026425612160354146,
+      "loss": 0.0082,
+      "step": 5700
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642435604613012,
+      "loss": 0.0095,
+      "step": 5701
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264230997410978,
+      "loss": 0.0054,
+      "step": 5702
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642184324527814,
+      "loss": 0.0008,
+      "step": 5703
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642058655869214,
+      "loss": 0.0182,
+      "step": 5704
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026419329681360784,
+      "loss": 0.012,
+      "step": 5705
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641807261330507,
+      "loss": 0.0058,
+      "step": 5706
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641681535454599,
+      "loss": 0.007,
+      "step": 5707
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641555790510454,
+      "loss": 0.0357,
+      "step": 5708
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641430026500173,
+      "loss": 0.0062,
+      "step": 5709
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026413042434258555,
+      "loss": 0.0103,
+      "step": 5710
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026411784412896027,
+      "loss": 0.0254,
+      "step": 5711
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641052620093516,
+      "loss": 0.0039,
+      "step": 5712
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640926779839697,
+      "loss": 0.0159,
+      "step": 5713
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640800920530247,
+      "loss": 0.0281,
+      "step": 5714
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026406750421672674,
+      "loss": 0.0094,
+      "step": 5715
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640549144752862,
+      "loss": 0.0043,
+      "step": 5716
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026404232282891326,
+      "loss": 0.0195,
+      "step": 5717
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026402972927781835,
+      "loss": 0.0316,
+      "step": 5718
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026401713382221163,
+      "loss": 0.0155,
+      "step": 5719
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640045364623036,
+      "loss": 0.0087,
+      "step": 5720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026399193719830457,
+      "loss": 0.007,
+      "step": 5721
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026397933603042507,
+      "loss": 0.0279,
+      "step": 5722
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639667329588755,
+      "loss": 0.0134,
+      "step": 5723
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639541279838663,
+      "loss": 0.0249,
+      "step": 5724
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639415211056081,
+      "loss": 0.0229,
+      "step": 5725
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026392891232431145,
+      "loss": 0.0033,
+      "step": 5726
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026391630164018684,
+      "loss": 0.0093,
+      "step": 5727
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000263903689053445,
+      "loss": 0.0108,
+      "step": 5728
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638910745642965,
+      "loss": 0.0411,
+      "step": 5729
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638784581729521,
+      "loss": 0.01,
+      "step": 5730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638658398796225,
+      "loss": 0.0051,
+      "step": 5731
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638532196845185,
+      "loss": 0.03,
+      "step": 5732
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638405975878507,
+      "loss": 0.007,
+      "step": 5733
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638279735898301,
+      "loss": 0.0286,
+      "step": 5734
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026381534769066747,
+      "loss": 0.0294,
+      "step": 5735
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638027198905737,
+      "loss": 0.0212,
+      "step": 5736
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026379009018975965,
+      "loss": 0.0218,
+      "step": 5737
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002637774585884363,
+      "loss": 0.0218,
+      "step": 5738
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002637648250868146,
+      "loss": 0.0134,
+      "step": 5739
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002637521896851057,
+      "loss": 0.0064,
+      "step": 5740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026373955238352036,
+      "loss": 0.0347,
+      "step": 5741
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026372691318226986,
+      "loss": 0.0152,
+      "step": 5742
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026371427208156525,
+      "loss": 0.0253,
+      "step": 5743
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026370162908161766,
+      "loss": 0.0084,
+      "step": 5744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026368898418263815,
+      "loss": 0.0083,
+      "step": 5745
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026367633738483807,
+      "loss": 0.0318,
+      "step": 5746
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002636636886884286,
+      "loss": 0.0182,
+      "step": 5747
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002636510380936209,
+      "loss": 0.0171,
+      "step": 5748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002636383856006264,
+      "loss": 0.0284,
+      "step": 5749
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002636257312096563,
+      "loss": 0.0274,
+      "step": 5750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026361307492092206,
+      "loss": 0.0124,
+      "step": 5751
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026360041673463493,
+      "loss": 0.0172,
+      "step": 5752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026358775665100647,
+      "loss": 0.0102,
+      "step": 5753
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002635750946702481,
+      "loss": 0.0086,
+      "step": 5754
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026356243079257117,
+      "loss": 0.0286,
+      "step": 5755
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026354976501818733,
+      "loss": 0.0213,
+      "step": 5756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026353709734730805,
+      "loss": 0.0103,
+      "step": 5757
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002635244277801449,
+      "loss": 0.0089,
+      "step": 5758
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026351175631690956,
+      "loss": 0.0188,
+      "step": 5759
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002634990829578136,
+      "loss": 0.0328,
+      "step": 5760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026348640770306874,
+      "loss": 0.0169,
+      "step": 5761
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026347373055288665,
+      "loss": 0.008,
+      "step": 5762
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026346105150747904,
+      "loss": 0.0245,
+      "step": 5763
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002634483705670577,
+      "loss": 0.0282,
+      "step": 5764
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026343568773183444,
+      "loss": 0.0298,
+      "step": 5765
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026342300300202104,
+      "loss": 0.0211,
+      "step": 5766
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002634103163778294,
+      "loss": 0.0133,
+      "step": 5767
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002633976278594714,
+      "loss": 0.0066,
+      "step": 5768
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000263384937447159,
+      "loss": 0.0388,
+      "step": 5769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026337224514110403,
+      "loss": 0.0065,
+      "step": 5770
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026335955094151865,
+      "loss": 0.0178,
+      "step": 5771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026334685484861477,
+      "loss": 0.0252,
+      "step": 5772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002633341568626044,
+      "loss": 0.0215,
+      "step": 5773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002633214569836997,
+      "loss": 0.0247,
+      "step": 5774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002633087552121128,
+      "loss": 0.0515,
+      "step": 5775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026329605154805575,
+      "loss": 0.0032,
+      "step": 5776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026328334599174085,
+      "loss": 0.0063,
+      "step": 5777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026327063854338014,
+      "loss": 0.0192,
+      "step": 5778
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000263257929203186,
+      "loss": 0.0108,
+      "step": 5779
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026324521797137067,
+      "loss": 0.0194,
+      "step": 5780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632325048481464,
+      "loss": 0.0311,
+      "step": 5781
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632197898337256,
+      "loss": 0.0101,
+      "step": 5782
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026320707292832056,
+      "loss": 0.0206,
+      "step": 5783
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002631943541321437,
+      "loss": 0.0281,
+      "step": 5784
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002631816334454075,
+      "loss": 0.0083,
+      "step": 5785
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026316891086832433,
+      "loss": 0.0094,
+      "step": 5786
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026315618640110675,
+      "loss": 0.0108,
+      "step": 5787
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026314346004396725,
+      "loss": 0.0231,
+      "step": 5788
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026313073179711844,
+      "loss": 0.0107,
+      "step": 5789
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026311800166077277,
+      "loss": 0.0052,
+      "step": 5790
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000263105269635143,
+      "loss": 0.017,
+      "step": 5791
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026309253572044173,
+      "loss": 0.017,
+      "step": 5792
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026307979991688164,
+      "loss": 0.0259,
+      "step": 5793
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630670622246754,
+      "loss": 0.0112,
+      "step": 5794
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630543226440358,
+      "loss": 0.0089,
+      "step": 5795
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026304158117517563,
+      "loss": 0.0195,
+      "step": 5796
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630288378183077,
+      "loss": 0.0098,
+      "step": 5797
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630160925736448,
+      "loss": 0.0206,
+      "step": 5798
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026300334544139976,
+      "loss": 0.022,
+      "step": 5799
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629905964217856,
+      "loss": 0.04,
+      "step": 5800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026297784551501517,
+      "loss": 0.0213,
+      "step": 5801
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629650927213015,
+      "loss": 0.0334,
+      "step": 5802
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629523380408575,
+      "loss": 0.0144,
+      "step": 5803
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629395814738963,
+      "loss": 0.0031,
+      "step": 5804
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026292682302063084,
+      "loss": 0.0199,
+      "step": 5805
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629140626812743,
+      "loss": 0.0218,
+      "step": 5806
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629013004560398,
+      "loss": 0.0092,
+      "step": 5807
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628885363451404,
+      "loss": 0.0261,
+      "step": 5808
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628757703487894,
+      "loss": 0.0097,
+      "step": 5809
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026286300246719995,
+      "loss": 0.0097,
+      "step": 5810
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628502327005853,
+      "loss": 0.0276,
+      "step": 5811
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628374610491588,
+      "loss": 0.0042,
+      "step": 5812
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628246875131336,
+      "loss": 0.0064,
+      "step": 5813
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026281191209272324,
+      "loss": 0.0075,
+      "step": 5814
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026279913478814096,
+      "loss": 0.0121,
+      "step": 5815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002627863555996002,
+      "loss": 0.0239,
+      "step": 5816
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026277357452731445,
+      "loss": 0.0312,
+      "step": 5817
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002627607915714971,
+      "loss": 0.0126,
+      "step": 5818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026274800673236167,
+      "loss": 0.0255,
+      "step": 5819
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002627352200101217,
+      "loss": 0.0118,
+      "step": 5820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002627224314049908,
+      "loss": 0.0235,
+      "step": 5821
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026270964091718247,
+      "loss": 0.0388,
+      "step": 5822
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026269684854691034,
+      "loss": 0.0043,
+      "step": 5823
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026268405429438817,
+      "loss": 0.0116,
+      "step": 5824
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026267125815982963,
+      "loss": 0.0189,
+      "step": 5825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026265846014344833,
+      "loss": 0.0109,
+      "step": 5826
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002626456602454581,
+      "loss": 0.003,
+      "step": 5827
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026263285846607264,
+      "loss": 0.0185,
+      "step": 5828
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262620054805506,
+      "loss": 0.0095,
+      "step": 5829
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026260724926397176,
+      "loss": 0.0025,
+      "step": 5830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026259444184168385,
+      "loss": 0.0211,
+      "step": 5831
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002625816325388563,
+      "loss": 0.0126,
+      "step": 5832
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026256882135570293,
+      "loss": 0.0187,
+      "step": 5833
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002625560082924378,
+      "loss": 0.0048,
+      "step": 5834
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002625431933492748,
+      "loss": 0.0022,
+      "step": 5835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026253037652642806,
+      "loss": 0.0023,
+      "step": 5836
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026251755782411164,
+      "loss": 0.0044,
+      "step": 5837
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026250473724253963,
+      "loss": 0.0158,
+      "step": 5838
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624919147819261,
+      "loss": 0.0096,
+      "step": 5839
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624790904424853,
+      "loss": 0.0204,
+      "step": 5840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624662642244313,
+      "loss": 0.0111,
+      "step": 5841
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624534361279784,
+      "loss": 0.0054,
+      "step": 5842
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624406061533409,
+      "loss": 0.0033,
+      "step": 5843
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262427774300733,
+      "loss": 0.0414,
+      "step": 5844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262414940570369,
+      "loss": 0.0247,
+      "step": 5845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624021049624634,
+      "loss": 0.0242,
+      "step": 5846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026238926747723044,
+      "loss": 0.0075,
+      "step": 5847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002623764281148845,
+      "loss": 0.0197,
+      "step": 5848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002623635868756401,
+      "loss": 0.0124,
+      "step": 5849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026235074375971176,
+      "loss": 0.0033,
+      "step": 5850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026233789876731385,
+      "loss": 0.0252,
+      "step": 5851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262325051898661,
+      "loss": 0.0064,
+      "step": 5852
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002623122031539678,
+      "loss": 0.0103,
+      "step": 5853
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026229935253344874,
+      "loss": 0.0152,
+      "step": 5854
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026228650003731855,
+      "loss": 0.0077,
+      "step": 5855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622736456657918,
+      "loss": 0.03,
+      "step": 5856
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622607894190833,
+      "loss": 0.0127,
+      "step": 5857
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622479312974077,
+      "loss": 0.0318,
+      "step": 5858
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026223507130097975,
+      "loss": 0.0111,
+      "step": 5859
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026222220943001424,
+      "loss": 0.0057,
+      "step": 5860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262209345684726,
+      "loss": 0.0038,
+      "step": 5861
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621964800653299,
+      "loss": 0.004,
+      "step": 5862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621836125720408,
+      "loss": 0.0132,
+      "step": 5863
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621707432050736,
+      "loss": 0.0083,
+      "step": 5864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026215787196464323,
+      "loss": 0.0219,
+      "step": 5865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026214499885096474,
+      "loss": 0.0089,
+      "step": 5866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262132123864253,
+      "loss": 0.0194,
+      "step": 5867
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621192470047232,
+      "loss": 0.0498,
+      "step": 5868
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002621063682725903,
+      "loss": 0.0147,
+      "step": 5869
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620934876680694,
+      "loss": 0.0158,
+      "step": 5870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026208060519137573,
+      "loss": 0.0055,
+      "step": 5871
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620677208427243,
+      "loss": 0.0079,
+      "step": 5872
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620548346223305,
+      "loss": 0.0185,
+      "step": 5873
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620419465304094,
+      "loss": 0.0323,
+      "step": 5874
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620290565671763,
+      "loss": 0.0048,
+      "step": 5875
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026201616473284646,
+      "loss": 0.0084,
+      "step": 5876
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620032710276352,
+      "loss": 0.0192,
+      "step": 5877
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026199037545175793,
+      "loss": 0.0072,
+      "step": 5878
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026197747800543004,
+      "loss": 0.0312,
+      "step": 5879
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026196457868886685,
+      "loss": 0.0268,
+      "step": 5880
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026195167750228383,
+      "loss": 0.0058,
+      "step": 5881
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619387744458965,
+      "loss": 0.0141,
+      "step": 5882
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619258695199203,
+      "loss": 0.0295,
+      "step": 5883
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026191296272457085,
+      "loss": 0.034,
+      "step": 5884
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026190005406006363,
+      "loss": 0.0162,
+      "step": 5885
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618871435266143,
+      "loss": 0.0077,
+      "step": 5886
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618742311244385,
+      "loss": 0.0068,
+      "step": 5887
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026186131685375185,
+      "loss": 0.0431,
+      "step": 5888
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026184840071477,
+      "loss": 0.0062,
+      "step": 5889
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618354827077088,
+      "loss": 0.0376,
+      "step": 5890
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618225628327839,
+      "loss": 0.0298,
+      "step": 5891
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618096410902111,
+      "loss": 0.0128,
+      "step": 5892
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002617967174802063,
+      "loss": 0.0215,
+      "step": 5893
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002617837920029852,
+      "loss": 0.0063,
+      "step": 5894
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026177086465876384,
+      "loss": 0.0013,
+      "step": 5895
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026175793544775804,
+      "loss": 0.0353,
+      "step": 5896
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026174500437018376,
+      "loss": 0.019,
+      "step": 5897
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000261732071426257,
+      "loss": 0.0042,
+      "step": 5898
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026171913661619366,
+      "loss": 0.0178,
+      "step": 5899
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002617061999402099,
+      "loss": 0.0054,
+      "step": 5900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026169326139852177,
+      "loss": 0.0161,
+      "step": 5901
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616803209913453,
+      "loss": 0.026,
+      "step": 5902
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026166737871889667,
+      "loss": 0.0102,
+      "step": 5903
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000261654434581392,
+      "loss": 0.0145,
+      "step": 5904
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026164148857904756,
+      "loss": 0.0147,
+      "step": 5905
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616285407120795,
+      "loss": 0.033,
+      "step": 5906
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616155909807041,
+      "loss": 0.0122,
+      "step": 5907
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616026393851376,
+      "loss": 0.0141,
+      "step": 5908
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002615896859255964,
+      "loss": 0.0089,
+      "step": 5909
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026157673060229683,
+      "loss": 0.0195,
+      "step": 5910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002615637734154552,
+      "loss": 0.0231,
+      "step": 5911
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000261550814365288,
+      "loss": 0.0224,
+      "step": 5912
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026153785345201154,
+      "loss": 0.0198,
+      "step": 5913
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026152489067584245,
+      "loss": 0.0237,
+      "step": 5914
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026151192603699716,
+      "loss": 0.0085,
+      "step": 5915
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026149895953569226,
+      "loss": 0.0247,
+      "step": 5916
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002614859911721442,
+      "loss": 0.0176,
+      "step": 5917
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026147302094656964,
+      "loss": 0.0332,
+      "step": 5918
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614600488591853,
+      "loss": 0.0214,
+      "step": 5919
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614470749102077,
+      "loss": 0.0095,
+      "step": 5920
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026143409909985354,
+      "loss": 0.0227,
+      "step": 5921
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614211214283396,
+      "loss": 0.0192,
+      "step": 5922
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614081418958826,
+      "loss": 0.036,
+      "step": 5923
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613951605026994,
+      "loss": 0.0277,
+      "step": 5924
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026138217724900665,
+      "loss": 0.0283,
+      "step": 5925
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026136919213502134,
+      "loss": 0.0236,
+      "step": 5926
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613562051609603,
+      "loss": 0.022,
+      "step": 5927
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613432163270404,
+      "loss": 0.0243,
+      "step": 5928
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026133022563347864,
+      "loss": 0.0147,
+      "step": 5929
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000261317233080492,
+      "loss": 0.0112,
+      "step": 5930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613042386682974,
+      "loss": 0.0061,
+      "step": 5931
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026129124239711185,
+      "loss": 0.0123,
+      "step": 5932
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026127824426715255,
+      "loss": 0.011,
+      "step": 5933
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002612652442786365,
+      "loss": 0.021,
+      "step": 5934
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002612522424317808,
+      "loss": 0.0078,
+      "step": 5935
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002612392387268028,
+      "loss": 0.0269,
+      "step": 5936
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002612262331639194,
+      "loss": 0.0064,
+      "step": 5937
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026121322574334795,
+      "loss": 0.0063,
+      "step": 5938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026120021646530576,
+      "loss": 0.0086,
+      "step": 5939
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026118720533001,
+      "loss": 0.0076,
+      "step": 5940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026117419233767804,
+      "loss": 0.0048,
+      "step": 5941
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026116117748852723,
+      "loss": 0.008,
+      "step": 5942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026114816078277495,
+      "loss": 0.0105,
+      "step": 5943
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026113514222063854,
+      "loss": 0.0137,
+      "step": 5944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026112212180233544,
+      "loss": 0.0064,
+      "step": 5945
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002611090995280832,
+      "loss": 0.0202,
+      "step": 5946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026109607539809927,
+      "loss": 0.0082,
+      "step": 5947
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002610830494126012,
+      "loss": 0.0191,
+      "step": 5948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002610700215718064,
+      "loss": 0.021,
+      "step": 5949
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026105699187593264,
+      "loss": 0.0113,
+      "step": 5950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026104396032519754,
+      "loss": 0.0089,
+      "step": 5951
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026103092691981866,
+      "loss": 0.0092,
+      "step": 5952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026101789166001367,
+      "loss": 0.0048,
+      "step": 5953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002610048545460004,
+      "loss": 0.0227,
+      "step": 5954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026099181557799644,
+      "loss": 0.003,
+      "step": 5955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002609787747562197,
+      "loss": 0.001,
+      "step": 5956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026096573208088793,
+      "loss": 0.0031,
+      "step": 5957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026095268755221894,
+      "loss": 0.0202,
+      "step": 5958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002609396411704307,
+      "loss": 0.0015,
+      "step": 5959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026092659293574095,
+      "loss": 0.0123,
+      "step": 5960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026091354284836775,
+      "loss": 0.0094,
+      "step": 5961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026090049090852905,
+      "loss": 0.0237,
+      "step": 5962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026088743711644277,
+      "loss": 0.0243,
+      "step": 5963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000260874381472327,
+      "loss": 0.0308,
+      "step": 5964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002608613239763997,
+      "loss": 0.0363,
+      "step": 5965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026084826462887913,
+      "loss": 0.0168,
+      "step": 5966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026083520342998326,
+      "loss": 0.0034,
+      "step": 5967
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002608221403799303,
+      "loss": 0.015,
+      "step": 5968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026080907547893834,
+      "loss": 0.0114,
+      "step": 5969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607960087272257,
+      "loss": 0.0118,
+      "step": 5970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026078294012501054,
+      "loss": 0.0082,
+      "step": 5971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026076986967251125,
+      "loss": 0.0275,
+      "step": 5972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026075679736994596,
+      "loss": 0.0161,
+      "step": 5973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607437232175331,
+      "loss": 0.0182,
+      "step": 5974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000260730647215491,
+      "loss": 0.0111,
+      "step": 5975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026071756936403806,
+      "loss": 0.0138,
+      "step": 5976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607044896633928,
+      "loss": 0.0072,
+      "step": 5977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026069140811377353,
+      "loss": 0.0148,
+      "step": 5978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606783247153988,
+      "loss": 0.0269,
+      "step": 5979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606652394684871,
+      "loss": 0.0131,
+      "step": 5980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606521523732571,
+      "loss": 0.0516,
+      "step": 5981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026063906342992723,
+      "loss": 0.0468,
+      "step": 5982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606259726387161,
+      "loss": 0.0375,
+      "step": 5983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026061287999984255,
+      "loss": 0.0159,
+      "step": 5984
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026059978551352497,
+      "loss": 0.0133,
+      "step": 5985
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605866891799822,
+      "loss": 0.0077,
+      "step": 5986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605735909994331,
+      "loss": 0.0114,
+      "step": 5987
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605604909720962,
+      "loss": 0.0239,
+      "step": 5988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026054738909819043,
+      "loss": 0.0111,
+      "step": 5989
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605342853779346,
+      "loss": 0.0163,
+      "step": 5990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026052117981154754,
+      "loss": 0.0166,
+      "step": 5991
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026050807239924817,
+      "loss": 0.014,
+      "step": 5992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604949631412554,
+      "loss": 0.0265,
+      "step": 5993
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604818520377881,
+      "loss": 0.0072,
+      "step": 5994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604687390890654,
+      "loss": 0.0118,
+      "step": 5995
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026045562429530625,
+      "loss": 0.0123,
+      "step": 5996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604425076567296,
+      "loss": 0.0209,
+      "step": 5997
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026042938917355467,
+      "loss": 0.0188,
+      "step": 5998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604162688460004,
+      "loss": 0.0295,
+      "step": 5999
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604031466742861,
+      "loss": 0.0181,
+      "step": 6000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026039002265863077,
+      "loss": 0.0154,
+      "step": 6001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002603768967992537,
+      "loss": 0.0432,
+      "step": 6002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002603637690963742,
+      "loss": 0.0156,
+      "step": 6003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026035063955021136,
+      "loss": 0.0377,
+      "step": 6004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026033750816098456,
+      "loss": 0.0182,
+      "step": 6005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026032437492891304,
+      "loss": 0.0059,
+      "step": 6006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026031123985421624,
+      "loss": 0.0138,
+      "step": 6007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602981029371135,
+      "loss": 0.0257,
+      "step": 6008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602849641778243,
+      "loss": 0.0088,
+      "step": 6009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026027182357656796,
+      "loss": 0.0176,
+      "step": 6010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000260258681133564,
+      "loss": 0.0182,
+      "step": 6011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026024553684903196,
+      "loss": 0.0086,
+      "step": 6012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602323907231914,
+      "loss": 0.0116,
+      "step": 6013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026021924275626177,
+      "loss": 0.0043,
+      "step": 6014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026020609294846275,
+      "loss": 0.0158,
+      "step": 6015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002601929413000139,
+      "loss": 0.0119,
+      "step": 6016
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026017978781113504,
+      "loss": 0.0041,
+      "step": 6017
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002601666324820457,
+      "loss": 0.0092,
+      "step": 6018
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002601534753129656,
+      "loss": 0.016,
+      "step": 6019
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026014031630411453,
+      "loss": 0.0083,
+      "step": 6020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026012715545571235,
+      "loss": 0.0054,
+      "step": 6021
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002601139927679787,
+      "loss": 0.0391,
+      "step": 6022
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002601008282411336,
+      "loss": 0.0028,
+      "step": 6023
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026008766187539677,
+      "loss": 0.0067,
+      "step": 6024
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600744936709882,
+      "loss": 0.0128,
+      "step": 6025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026006132362812784,
+      "loss": 0.0292,
+      "step": 6026
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600481517470356,
+      "loss": 0.0112,
+      "step": 6027
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026003497802793145,
+      "loss": 0.0028,
+      "step": 6028
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026002180247103557,
+      "loss": 0.0323,
+      "step": 6029
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600086250765678,
+      "loss": 0.0109,
+      "step": 6030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002599954458447484,
+      "loss": 0.0329,
+      "step": 6031
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025998226477579734,
+      "loss": 0.0137,
+      "step": 6032
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025996908186993496,
+      "loss": 0.0148,
+      "step": 6033
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025995589712738125,
+      "loss": 0.0051,
+      "step": 6034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002599427105483566,
+      "loss": 0.005,
+      "step": 6035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000259929522133081,
+      "loss": 0.0147,
+      "step": 6036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000259916331881775,
+      "loss": 0.0196,
+      "step": 6037
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002599031397946587,
+      "loss": 0.0298,
+      "step": 6038
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598899458719526,
+      "loss": 0.0146,
+      "step": 6039
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025987675011387686,
+      "loss": 0.0102,
+      "step": 6040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025986355252065205,
+      "loss": 0.0122,
+      "step": 6041
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598503530924986,
+      "loss": 0.012,
+      "step": 6042
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025983715182963677,
+      "loss": 0.0153,
+      "step": 6043
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025982394873228724,
+      "loss": 0.0173,
+      "step": 6044
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598107438006704,
+      "loss": 0.0195,
+      "step": 6045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025979753703500694,
+      "loss": 0.0158,
+      "step": 6046
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025978432843551727,
+      "loss": 0.0076,
+      "step": 6047
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025977111800242213,
+      "loss": 0.0238,
+      "step": 6048
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597579057359421,
+      "loss": 0.0113,
+      "step": 6049
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597446916362979,
+      "loss": 0.0251,
+      "step": 6050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025973147570371017,
+      "loss": 0.0068,
+      "step": 6051
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597182579383996,
+      "loss": 0.014,
+      "step": 6052
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597050383405871,
+      "loss": 0.0114,
+      "step": 6053
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025969181691049334,
+      "loss": 0.0113,
+      "step": 6054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025967859364833915,
+      "loss": 0.0258,
+      "step": 6055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596653685543455,
+      "loss": 0.0241,
+      "step": 6056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025965214162873313,
+      "loss": 0.0399,
+      "step": 6057
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025963891287172295,
+      "loss": 0.0109,
+      "step": 6058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596256822835361,
+      "loss": 0.0118,
+      "step": 6059
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596124498643933,
+      "loss": 0.0065,
+      "step": 6060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002595992156145157,
+      "loss": 0.0053,
+      "step": 6061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002595859795341244,
+      "loss": 0.009,
+      "step": 6062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025957274162344035,
+      "loss": 0.0197,
+      "step": 6063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025955950188268464,
+      "loss": 0.0175,
+      "step": 6064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025954626031207846,
+      "loss": 0.0049,
+      "step": 6065
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025953301691184296,
+      "loss": 0.0215,
+      "step": 6066
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002595197716821993,
+      "loss": 0.0043,
+      "step": 6067
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002595065246233688,
+      "loss": 0.0067,
+      "step": 6068
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594932757355725,
+      "loss": 0.0101,
+      "step": 6069
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594800250190319,
+      "loss": 0.0341,
+      "step": 6070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025946677247396817,
+      "loss": 0.0197,
+      "step": 6071
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025945351810060277,
+      "loss": 0.0151,
+      "step": 6072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594402618991569,
+      "loss": 0.0059,
+      "step": 6073
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594270038698522,
+      "loss": 0.0206,
+      "step": 6074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025941374401290993,
+      "loss": 0.0019,
+      "step": 6075
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025940048232855163,
+      "loss": 0.0136,
+      "step": 6076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593872188169987,
+      "loss": 0.0397,
+      "step": 6077
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593739534784728,
+      "loss": 0.0082,
+      "step": 6078
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025936068631319533,
+      "loss": 0.0085,
+      "step": 6079
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000259347417321388,
+      "loss": 0.0423,
+      "step": 6080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025933414650327247,
+      "loss": 0.0178,
+      "step": 6081
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025932087385907023,
+      "loss": 0.0066,
+      "step": 6082
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593075993890031,
+      "loss": 0.0276,
+      "step": 6083
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025929432309329265,
+      "loss": 0.0066,
+      "step": 6084
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025928104497216075,
+      "loss": 0.003,
+      "step": 6085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592677650258291,
+      "loss": 0.0091,
+      "step": 6086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592544832545195,
+      "loss": 0.0295,
+      "step": 6087
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025924119965845384,
+      "loss": 0.0386,
+      "step": 6088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025922791423785394,
+      "loss": 0.0078,
+      "step": 6089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592146269929417,
+      "loss": 0.0011,
+      "step": 6090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025920133792393897,
+      "loss": 0.0055,
+      "step": 6091
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002591880470310678,
+      "loss": 0.0236,
+      "step": 6092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025917475431455015,
+      "loss": 0.0208,
+      "step": 6093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000259161459774608,
+      "loss": 0.0202,
+      "step": 6094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025914816341146335,
+      "loss": 0.0115,
+      "step": 6095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002591348652253384,
+      "loss": 0.022,
+      "step": 6096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025912156521645525,
+      "loss": 0.0161,
+      "step": 6097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002591082633850359,
+      "loss": 0.0149,
+      "step": 6098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590949597313027,
+      "loss": 0.0246,
+      "step": 6099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025908165425547763,
+      "loss": 0.0069,
+      "step": 6100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000259068346957783,
+      "loss": 0.0055,
+      "step": 6101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590550378384412,
+      "loss": 0.0023,
+      "step": 6102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590417268976743,
+      "loss": 0.0101,
+      "step": 6103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025902841413570486,
+      "loss": 0.0132,
+      "step": 6104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000259015099552755,
+      "loss": 0.0069,
+      "step": 6105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025900178314904726,
+      "loss": 0.0324,
+      "step": 6106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000258988464924804,
+      "loss": 0.0258,
+      "step": 6107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589751448802476,
+      "loss": 0.092,
+      "step": 6108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589618230156006,
+      "loss": 0.0247,
+      "step": 6109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025894849933108543,
+      "loss": 0.0127,
+      "step": 6110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589351738269247,
+      "loss": 0.0019,
+      "step": 6111
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025892184650334096,
+      "loss": 0.0249,
+      "step": 6112
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589085173605567,
+      "loss": 0.0214,
+      "step": 6113
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025889518639879465,
+      "loss": 0.0068,
+      "step": 6114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002588818536182775,
+      "loss": 0.0103,
+      "step": 6115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002588685190192278,
+      "loss": 0.0087,
+      "step": 6116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025885518260186836,
+      "loss": 0.0153,
+      "step": 6117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002588418443664219,
+      "loss": 0.0238,
+      "step": 6118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025882850431311114,
+      "loss": 0.024,
+      "step": 6119
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002588151624421589,
+      "loss": 0.0026,
+      "step": 6120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025880181875378815,
+      "loss": 0.0145,
+      "step": 6121
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025878847324822157,
+      "loss": 0.0037,
+      "step": 6122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587751259256822,
+      "loss": 0.0036,
+      "step": 6123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587617767863929,
+      "loss": 0.0066,
+      "step": 6124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587484258305766,
+      "loss": 0.0142,
+      "step": 6125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025873507305845627,
+      "loss": 0.0183,
+      "step": 6126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000258721718470255,
+      "loss": 0.0363,
+      "step": 6127
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587083620661958,
+      "loss": 0.0105,
+      "step": 6128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025869500384650176,
+      "loss": 0.0067,
+      "step": 6129
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025868164381139593,
+      "loss": 0.0175,
+      "step": 6130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025866828196110153,
+      "loss": 0.0037,
+      "step": 6131
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025865491829584175,
+      "loss": 0.0051,
+      "step": 6132
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025864155281583965,
+      "loss": 0.0121,
+      "step": 6133
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586281855213186,
+      "loss": 0.0207,
+      "step": 6134
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586148164125018,
+      "loss": 0.0174,
+      "step": 6135
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586014454896125,
+      "loss": 0.0242,
+      "step": 6136
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025858807275287406,
+      "loss": 0.0171,
+      "step": 6137
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585746982025098,
+      "loss": 0.0312,
+      "step": 6138
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585613218387433,
+      "loss": 0.0328,
+      "step": 6139
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585479436617976,
+      "loss": 0.0136,
+      "step": 6140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585345636718965,
+      "loss": 0.0038,
+      "step": 6141
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585211818692632,
+      "loss": 0.0088,
+      "step": 6142
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585077982541214,
+      "loss": 0.0073,
+      "step": 6143
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584944128266945,
+      "loss": 0.0106,
+      "step": 6144
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584810255872061,
+      "loss": 0.0082,
+      "step": 6145
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584676365358798,
+      "loss": 0.0105,
+      "step": 6146
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584542456729392,
+      "loss": 0.0097,
+      "step": 6147
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000258440852998608,
+      "loss": 0.0106,
+      "step": 6148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584274585131099,
+      "loss": 0.0072,
+      "step": 6149
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025841406221666855,
+      "loss": 0.005,
+      "step": 6150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025840066410950767,
+      "loss": 0.0088,
+      "step": 6151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002583872641918511,
+      "loss": 0.0038,
+      "step": 6152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025837386246392265,
+      "loss": 0.0168,
+      "step": 6153
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025836045892594614,
+      "loss": 0.0074,
+      "step": 6154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002583470535781454,
+      "loss": 0.0062,
+      "step": 6155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002583336464207443,
+      "loss": 0.0126,
+      "step": 6156
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025832023745396684,
+      "loss": 0.0288,
+      "step": 6157
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000258306826678037,
+      "loss": 0.0119,
+      "step": 6158
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582934140931786,
+      "loss": 0.0077,
+      "step": 6159
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582799996996159,
+      "loss": 0.0126,
+      "step": 6160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025826658349757274,
+      "loss": 0.0024,
+      "step": 6161
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025825316548727325,
+      "loss": 0.0081,
+      "step": 6162
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582397456689416,
+      "loss": 0.0198,
+      "step": 6163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025822632404280183,
+      "loss": 0.0232,
+      "step": 6164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025821290060907816,
+      "loss": 0.008,
+      "step": 6165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025819947536799483,
+      "loss": 0.0215,
+      "step": 6166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000258186048319776,
+      "loss": 0.039,
+      "step": 6167
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581726194646459,
+      "loss": 0.0234,
+      "step": 6168
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025815918880282894,
+      "loss": 0.0342,
+      "step": 6169
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025814575633454934,
+      "loss": 0.0039,
+      "step": 6170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581323220600314,
+      "loss": 0.0054,
+      "step": 6171
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025811888597949967,
+      "loss": 0.0112,
+      "step": 6172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025810544809317846,
+      "loss": 0.004,
+      "step": 6173
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025809200840129216,
+      "loss": 0.0166,
+      "step": 6174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025807856690406527,
+      "loss": 0.015,
+      "step": 6175
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580651236017223,
+      "loss": 0.0024,
+      "step": 6176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580516784944878,
+      "loss": 0.0294,
+      "step": 6177
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025803823158258626,
+      "loss": 0.0074,
+      "step": 6178
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025802478286624235,
+      "loss": 0.0212,
+      "step": 6179
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025801133234568065,
+      "loss": 0.0236,
+      "step": 6180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025799788002112584,
+      "loss": 0.0065,
+      "step": 6181
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002579844258928025,
+      "loss": 0.0316,
+      "step": 6182
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002579709699609355,
+      "loss": 0.0251,
+      "step": 6183
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025795751222574946,
+      "loss": 0.0188,
+      "step": 6184
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025794405268746916,
+      "loss": 0.045,
+      "step": 6185
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002579305913463195,
+      "loss": 0.0093,
+      "step": 6186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002579171282025251,
+      "loss": 0.0268,
+      "step": 6187
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025790366325631107,
+      "loss": 0.0144,
+      "step": 6188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578901965079021,
+      "loss": 0.01,
+      "step": 6189
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025787672795752324,
+      "loss": 0.0072,
+      "step": 6190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578632576053994,
+      "loss": 0.018,
+      "step": 6191
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578497854517555,
+      "loss": 0.0109,
+      "step": 6192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578363114968166,
+      "loss": 0.008,
+      "step": 6193
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578228357408078,
+      "loss": 0.0196,
+      "step": 6194
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000257809358183954,
+      "loss": 0.0114,
+      "step": 6195
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025779587882648056,
+      "loss": 0.0102,
+      "step": 6196
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577823976686124,
+      "loss": 0.012,
+      "step": 6197
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025776891471057477,
+      "loss": 0.0093,
+      "step": 6198
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577554299525928,
+      "loss": 0.0123,
+      "step": 6199
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577419433948918,
+      "loss": 0.009,
+      "step": 6200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577284550376969,
+      "loss": 0.0129,
+      "step": 6201
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577149648812335,
+      "loss": 0.0152,
+      "step": 6202
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025770147292572684,
+      "loss": 0.0137,
+      "step": 6203
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025768797917140226,
+      "loss": 0.019,
+      "step": 6204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576744836184851,
+      "loss": 0.024,
+      "step": 6205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025766098626720093,
+      "loss": 0.0058,
+      "step": 6206
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000257647487117775,
+      "loss": 0.0087,
+      "step": 6207
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576339861704329,
+      "loss": 0.0144,
+      "step": 6208
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576204834254,
+      "loss": 0.0274,
+      "step": 6209
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576069788829019,
+      "loss": 0.0218,
+      "step": 6210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025759347254316413,
+      "loss": 0.0076,
+      "step": 6211
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025757996440641223,
+      "loss": 0.0102,
+      "step": 6212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002575664544728719,
+      "loss": 0.0146,
+      "step": 6213
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002575529427427687,
+      "loss": 0.0133,
+      "step": 6214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002575394292163284,
+      "loss": 0.0168,
+      "step": 6215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025752591389377656,
+      "loss": 0.0221,
+      "step": 6216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025751239677533904,
+      "loss": 0.0147,
+      "step": 6217
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025749887786124154,
+      "loss": 0.0072,
+      "step": 6218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025748535715170985,
+      "loss": 0.0052,
+      "step": 6219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002574718346469698,
+      "loss": 0.0072,
+      "step": 6220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002574583103472472,
+      "loss": 0.0051,
+      "step": 6221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002574447842527681,
+      "loss": 0.0042,
+      "step": 6222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025743125636375815,
+      "loss": 0.0154,
+      "step": 6223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025741772668044354,
+      "loss": 0.0238,
+      "step": 6224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025740419520305004,
+      "loss": 0.0112,
+      "step": 6225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573906619318038,
+      "loss": 0.0056,
+      "step": 6226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025737712686693074,
+      "loss": 0.0136,
+      "step": 6227
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025736359000865704,
+      "loss": 0.0226,
+      "step": 6228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573500513572087,
+      "loss": 0.0307,
+      "step": 6229
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025733651091281183,
+      "loss": 0.0157,
+      "step": 6230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573229686756926,
+      "loss": 0.0145,
+      "step": 6231
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025730942464607725,
+      "loss": 0.0396,
+      "step": 6232
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025729587882419196,
+      "loss": 0.008,
+      "step": 6233
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025728233121026295,
+      "loss": 0.0092,
+      "step": 6234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572687818045165,
+      "loss": 0.0196,
+      "step": 6235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572552306071789,
+      "loss": 0.0098,
+      "step": 6236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572416776184765,
+      "loss": 0.016,
+      "step": 6237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572281228386356,
+      "loss": 0.009,
+      "step": 6238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572145662678827,
+      "loss": 0.0105,
+      "step": 6239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025720100790644414,
+      "loss": 0.0041,
+      "step": 6240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025718744775454636,
+      "loss": 0.0486,
+      "step": 6241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025717388581241596,
+      "loss": 0.0051,
+      "step": 6242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025716032208027926,
+      "loss": 0.0121,
+      "step": 6243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000257146756558363,
+      "loss": 0.0186,
+      "step": 6244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025713318924689354,
+      "loss": 0.0261,
+      "step": 6245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025711962014609766,
+      "loss": 0.019,
+      "step": 6246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025710604925620193,
+      "loss": 0.0123,
+      "step": 6247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025709247657743295,
+      "loss": 0.0188,
+      "step": 6248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025707890211001754,
+      "loss": 0.0079,
+      "step": 6249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570653258541823,
+      "loss": 0.0108,
+      "step": 6250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025705174781015397,
+      "loss": 0.0108,
+      "step": 6251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570381679781594,
+      "loss": 0.0025,
+      "step": 6252
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570245863584254,
+      "loss": 0.0108,
+      "step": 6253
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570110029511788,
+      "loss": 0.0182,
+      "step": 6254
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025699741775664646,
+      "loss": 0.0073,
+      "step": 6255
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002569838307750552,
+      "loss": 0.0118,
+      "step": 6256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025697024200663205,
+      "loss": 0.0239,
+      "step": 6257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025695665145160395,
+      "loss": 0.0187,
+      "step": 6258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002569430591101979,
+      "loss": 0.0205,
+      "step": 6259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025692946498264086,
+      "loss": 0.0175,
+      "step": 6260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025691586906915993,
+      "loss": 0.0377,
+      "step": 6261
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025690227136998214,
+      "loss": 0.0064,
+      "step": 6262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568886718853346,
+      "loss": 0.0046,
+      "step": 6263
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568750706154446,
+      "loss": 0.0251,
+      "step": 6264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568614675605391,
+      "loss": 0.0323,
+      "step": 6265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568478627208453,
+      "loss": 0.0143,
+      "step": 6266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568342560965906,
+      "loss": 0.0063,
+      "step": 6267
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568206476880022,
+      "loss": 0.0186,
+      "step": 6268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025680703749530727,
+      "loss": 0.0107,
+      "step": 6269
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025679342551873316,
+      "loss": 0.0129,
+      "step": 6270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025677981175850737,
+      "loss": 0.0013,
+      "step": 6271
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567661962148571,
+      "loss": 0.0269,
+      "step": 6272
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567525788880098,
+      "loss": 0.0144,
+      "step": 6273
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567389597781929,
+      "loss": 0.0197,
+      "step": 6274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256725338885634,
+      "loss": 0.0184,
+      "step": 6275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567117162105604,
+      "loss": 0.015,
+      "step": 6276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566980917531997,
+      "loss": 0.0098,
+      "step": 6277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025668446551377946,
+      "loss": 0.0082,
+      "step": 6278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566708374925273,
+      "loss": 0.0085,
+      "step": 6279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566572076896708,
+      "loss": 0.0093,
+      "step": 6280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566435761054376,
+      "loss": 0.0309,
+      "step": 6281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566299427400553,
+      "loss": 0.0042,
+      "step": 6282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566163075937517,
+      "loss": 0.0127,
+      "step": 6283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566026706667546,
+      "loss": 0.0187,
+      "step": 6284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025658903195929165,
+      "loss": 0.0024,
+      "step": 6285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565753914715906,
+      "loss": 0.0049,
+      "step": 6286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025656174920387937,
+      "loss": 0.0216,
+      "step": 6287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025654810515638577,
+      "loss": 0.0022,
+      "step": 6288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025653445932933774,
+      "loss": 0.0211,
+      "step": 6289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565208117229631,
+      "loss": 0.0335,
+      "step": 6290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025650716233748986,
+      "loss": 0.0154,
+      "step": 6291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564935111731459,
+      "loss": 0.0151,
+      "step": 6292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564798582301593,
+      "loss": 0.0095,
+      "step": 6293
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025646620350875814,
+      "loss": 0.0098,
+      "step": 6294
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025645254700917033,
+      "loss": 0.015,
+      "step": 6295
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564388887316241,
+      "loss": 0.0045,
+      "step": 6296
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025642522867634746,
+      "loss": 0.0102,
+      "step": 6297
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564115668435687,
+      "loss": 0.0115,
+      "step": 6298
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563979032335158,
+      "loss": 0.0051,
+      "step": 6299
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563842378464171,
+      "loss": 0.0214,
+      "step": 6300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025637057068250084,
+      "loss": 0.032,
+      "step": 6301
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025635690174199527,
+      "loss": 0.0422,
+      "step": 6302
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563432310251287,
+      "loss": 0.0169,
+      "step": 6303
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025632955853212937,
+      "loss": 0.0091,
+      "step": 6304
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025631588426322575,
+      "loss": 0.0253,
+      "step": 6305
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025630220821864613,
+      "loss": 0.0038,
+      "step": 6306
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256288530398619,
+      "loss": 0.0085,
+      "step": 6307
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025627485080337276,
+      "loss": 0.0258,
+      "step": 6308
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562611694331359,
+      "loss": 0.0116,
+      "step": 6309
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256247486288137,
+      "loss": 0.0062,
+      "step": 6310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002562338013686045,
+      "loss": 0.0049,
+      "step": 6311
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002562201146747669,
+      "loss": 0.0201,
+      "step": 6312
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025620642620685294,
+      "loss": 0.0201,
+      "step": 6313
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002561927359650912,
+      "loss": 0.0319,
+      "step": 6314
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002561790439497103,
+      "loss": 0.0097,
+      "step": 6315
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025616535016093897,
+      "loss": 0.0051,
+      "step": 6316
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025615165459900584,
+      "loss": 0.0085,
+      "step": 6317
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025613795726413973,
+      "loss": 0.011,
+      "step": 6318
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025612425815656935,
+      "loss": 0.0213,
+      "step": 6319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002561105572765235,
+      "loss": 0.0363,
+      "step": 6320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560968546242312,
+      "loss": 0.0138,
+      "step": 6321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000256083150199921,
+      "loss": 0.0112,
+      "step": 6322
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025606944400382203,
+      "loss": 0.0246,
+      "step": 6323
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560557360361631,
+      "loss": 0.0117,
+      "step": 6324
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560420262971732,
+      "loss": 0.0066,
+      "step": 6325
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025602831478708127,
+      "loss": 0.0068,
+      "step": 6326
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560146015061164,
+      "loss": 0.0098,
+      "step": 6327
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560008864545075,
+      "loss": 0.0069,
+      "step": 6328
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002559871696324837,
+      "loss": 0.015,
+      "step": 6329
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025597345104027414,
+      "loss": 0.0193,
+      "step": 6330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025595973067810786,
+      "loss": 0.0305,
+      "step": 6331
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025594600854621414,
+      "loss": 0.0189,
+      "step": 6332
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025593228464482204,
+      "loss": 0.0182,
+      "step": 6333
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002559185589741608,
+      "loss": 0.0129,
+      "step": 6334
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002559048315344597,
+      "loss": 0.0048,
+      "step": 6335
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025589110232594807,
+      "loss": 0.01,
+      "step": 6336
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002558773713488551,
+      "loss": 0.0165,
+      "step": 6337
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025586363860341017,
+      "loss": 0.0259,
+      "step": 6338
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025584990408984265,
+      "loss": 0.0156,
+      "step": 6339
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002558361678083819,
+      "loss": 0.0102,
+      "step": 6340
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025582242975925743,
+      "loss": 0.0094,
+      "step": 6341
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002558086899426986,
+      "loss": 0.0068,
+      "step": 6342
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557949483589349,
+      "loss": 0.014,
+      "step": 6343
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025578120500819587,
+      "loss": 0.0172,
+      "step": 6344
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000255767459890711,
+      "loss": 0.0209,
+      "step": 6345
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025575371300671,
+      "loss": 0.0171,
+      "step": 6346
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557399643564223,
+      "loss": 0.0203,
+      "step": 6347
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557262139400776,
+      "loss": 0.0227,
+      "step": 6348
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557124617579056,
+      "loss": 0.0064,
+      "step": 6349
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025569870781013586,
+      "loss": 0.0121,
+      "step": 6350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556849520969982,
+      "loss": 0.0052,
+      "step": 6351
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556711946187224,
+      "loss": 0.0351,
+      "step": 6352
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025565743537553815,
+      "loss": 0.0058,
+      "step": 6353
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025564367436767526,
+      "loss": 0.0246,
+      "step": 6354
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025562991159536365,
+      "loss": 0.0098,
+      "step": 6355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025561614705883306,
+      "loss": 0.0421,
+      "step": 6356
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556023807583135,
+      "loss": 0.0078,
+      "step": 6357
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025558861269403484,
+      "loss": 0.0078,
+      "step": 6358
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025557484286622696,
+      "loss": 0.0234,
+      "step": 6359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025556107127512004,
+      "loss": 0.0052,
+      "step": 6360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002555472979209439,
+      "loss": 0.0145,
+      "step": 6361
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025553352280392863,
+      "loss": 0.0066,
+      "step": 6362
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025551974592430437,
+      "loss": 0.0189,
+      "step": 6363
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025550596728230113,
+      "loss": 0.0125,
+      "step": 6364
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025549218687814914,
+      "loss": 0.0163,
+      "step": 6365
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002554784047120784,
+      "loss": 0.0233,
+      "step": 6366
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025546462078431925,
+      "loss": 0.0181,
+      "step": 6367
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002554508350951019,
+      "loss": 0.0337,
+      "step": 6368
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025543704764465646,
+      "loss": 0.0464,
+      "step": 6369
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002554232584332134,
+      "loss": 0.0088,
+      "step": 6370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025540946746100286,
+      "loss": 0.0085,
+      "step": 6371
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025539567472825526,
+      "loss": 0.0232,
+      "step": 6372
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002553818802352009,
+      "loss": 0.0144,
+      "step": 6373
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002553680839820703,
+      "loss": 0.0365,
+      "step": 6374
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025535428596909374,
+      "loss": 0.0129,
+      "step": 6375
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025534048619650174,
+      "loss": 0.0102,
+      "step": 6376
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025532668466452473,
+      "loss": 0.0115,
+      "step": 6377
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002553128813733934,
+      "loss": 0.0433,
+      "step": 6378
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025529907632333807,
+      "loss": 0.032,
+      "step": 6379
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025528526951458936,
+      "loss": 0.0034,
+      "step": 6380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255271460947378,
+      "loss": 0.045,
+      "step": 6381
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002552576506219345,
+      "loss": 0.0057,
+      "step": 6382
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002552438385384895,
+      "loss": 0.0059,
+      "step": 6383
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025523002469727375,
+      "loss": 0.0264,
+      "step": 6384
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025521620909851796,
+      "loss": 0.0035,
+      "step": 6385
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025520239174245293,
+      "loss": 0.0024,
+      "step": 6386
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025518857262930926,
+      "loss": 0.0071,
+      "step": 6387
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255174751759318,
+      "loss": 0.0082,
+      "step": 6388
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002551609291327098,
+      "loss": 0.0318,
+      "step": 6389
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002551471047497155,
+      "loss": 0.0145,
+      "step": 6390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025513327861056615,
+      "loss": 0.0299,
+      "step": 6391
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002551194507154926,
+      "loss": 0.0525,
+      "step": 6392
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025510562106472573,
+      "loss": 0.0272,
+      "step": 6393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550917896584967,
+      "loss": 0.0253,
+      "step": 6394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550779564970363,
+      "loss": 0.0201,
+      "step": 6395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550641215805757,
+      "loss": 0.0068,
+      "step": 6396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255050284909346,
+      "loss": 0.0058,
+      "step": 6397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550364464835782,
+      "loss": 0.0211,
+      "step": 6398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550226063035035,
+      "loss": 0.0206,
+      "step": 6399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255008764369353,
+      "loss": 0.006,
+      "step": 6400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549949206813579,
+      "loss": 0.0187,
+      "step": 6401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549810752397495,
+      "loss": 0.0286,
+      "step": 6402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025496722804475895,
+      "loss": 0.0162,
+      "step": 6403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549533790966175,
+      "loss": 0.0074,
+      "step": 6404
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025493952839555656,
+      "loss": 0.0138,
+      "step": 6405
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549256759418074,
+      "loss": 0.0106,
+      "step": 6406
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549118217356014,
+      "loss": 0.0077,
+      "step": 6407
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002548979657771699,
+      "loss": 0.034,
+      "step": 6408
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002548841080667444,
+      "loss": 0.0309,
+      "step": 6409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025487024860455635,
+      "loss": 0.0116,
+      "step": 6410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025485638739083716,
+      "loss": 0.0202,
+      "step": 6411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002548425244258183,
+      "loss": 0.0045,
+      "step": 6412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025482865970973147,
+      "loss": 0.0198,
+      "step": 6413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002548147932428081,
+      "loss": 0.0339,
+      "step": 6414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025480092502527986,
+      "loss": 0.0143,
+      "step": 6415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025478705505737833,
+      "loss": 0.0064,
+      "step": 6416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025477318333933516,
+      "loss": 0.0077,
+      "step": 6417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025475930987138204,
+      "loss": 0.012,
+      "step": 6418
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002547454346537507,
+      "loss": 0.0069,
+      "step": 6419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025473155768667295,
+      "loss": 0.0122,
+      "step": 6420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025471767897038047,
+      "loss": 0.0234,
+      "step": 6421
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002547037985051051,
+      "loss": 0.0102,
+      "step": 6422
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546899162910786,
+      "loss": 0.0096,
+      "step": 6423
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546760323285329,
+      "loss": 0.0189,
+      "step": 6424
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546621466176998,
+      "loss": 0.0047,
+      "step": 6425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546482591588114,
+      "loss": 0.0103,
+      "step": 6426
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546343699520995,
+      "loss": 0.0069,
+      "step": 6427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025462047899779607,
+      "loss": 0.0162,
+      "step": 6428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546065862961332,
+      "loss": 0.0054,
+      "step": 6429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025459269184734286,
+      "loss": 0.0349,
+      "step": 6430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002545787956516572,
+      "loss": 0.0029,
+      "step": 6431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002545648977093082,
+      "loss": 0.0053,
+      "step": 6432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025455099802052796,
+      "loss": 0.0219,
+      "step": 6433
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025453709658554874,
+      "loss": 0.0183,
+      "step": 6434
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025452319340460276,
+      "loss": 0.0448,
+      "step": 6435
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025450928847792204,
+      "loss": 0.0066,
+      "step": 6436
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025449538180573897,
+      "loss": 0.0183,
+      "step": 6437
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002544814733882858,
+      "loss": 0.0114,
+      "step": 6438
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025446756322579475,
+      "loss": 0.0122,
+      "step": 6439
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025445365131849824,
+      "loss": 0.005,
+      "step": 6440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025443973766662854,
+      "loss": 0.0479,
+      "step": 6441
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002544258222704181,
+      "loss": 0.0071,
+      "step": 6442
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025441190513009936,
+      "loss": 0.0245,
+      "step": 6443
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543979862459046,
+      "loss": 0.0112,
+      "step": 6444
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543840656180665,
+      "loss": 0.0139,
+      "step": 6445
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543701432468174,
+      "loss": 0.0153,
+      "step": 6446
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025435621913239,
+      "loss": 0.0063,
+      "step": 6447
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025434229327501664,
+      "loss": 0.0146,
+      "step": 6448
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025432836567493007,
+      "loss": 0.0119,
+      "step": 6449
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543144363323629,
+      "loss": 0.0224,
+      "step": 6450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543005052475477,
+      "loss": 0.0227,
+      "step": 6451
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025428657242071716,
+      "loss": 0.0262,
+      "step": 6452
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000254272637852104,
+      "loss": 0.0343,
+      "step": 6453
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025425870154194096,
+      "loss": 0.0169,
+      "step": 6454
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025424476349046084,
+      "loss": 0.0145,
+      "step": 6455
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002542308236978964,
+      "loss": 0.0218,
+      "step": 6456
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025421688216448044,
+      "loss": 0.0064,
+      "step": 6457
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002542029388904458,
+      "loss": 0.031,
+      "step": 6458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025418899387602534,
+      "loss": 0.0043,
+      "step": 6459
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541750471214521,
+      "loss": 0.0261,
+      "step": 6460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541610986269589,
+      "loss": 0.0072,
+      "step": 6461
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025414714839277873,
+      "loss": 0.0077,
+      "step": 6462
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541331964191446,
+      "loss": 0.0015,
+      "step": 6463
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025411924270628955,
+      "loss": 0.0136,
+      "step": 6464
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025410528725444654,
+      "loss": 0.014,
+      "step": 6465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540913300638488,
+      "loss": 0.0092,
+      "step": 6466
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540773711347293,
+      "loss": 0.0132,
+      "step": 6467
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540634104673213,
+      "loss": 0.0203,
+      "step": 6468
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540494480618578,
+      "loss": 0.02,
+      "step": 6469
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540354839185722,
+      "loss": 0.0154,
+      "step": 6470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025402151803769764,
+      "loss": 0.0216,
+      "step": 6471
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540075504194673,
+      "loss": 0.0142,
+      "step": 6472
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539935810641146,
+      "loss": 0.0066,
+      "step": 6473
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539796099718728,
+      "loss": 0.0138,
+      "step": 6474
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025396563714297525,
+      "loss": 0.0067,
+      "step": 6475
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025395166257765527,
+      "loss": 0.0279,
+      "step": 6476
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025393768627614635,
+      "loss": 0.0224,
+      "step": 6477
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539237082386818,
+      "loss": 0.0124,
+      "step": 6478
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539097284654952,
+      "loss": 0.0153,
+      "step": 6479
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025389574695682004,
+      "loss": 0.0142,
+      "step": 6480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002538817637128897,
+      "loss": 0.005,
+      "step": 6481
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025386777873393786,
+      "loss": 0.0065,
+      "step": 6482
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000253853792020198,
+      "loss": 0.0122,
+      "step": 6483
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025383980357190385,
+      "loss": 0.0078,
+      "step": 6484
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002538258133892889,
+      "loss": 0.0109,
+      "step": 6485
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025381182147258695,
+      "loss": 0.016,
+      "step": 6486
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025379782782203157,
+      "loss": 0.0205,
+      "step": 6487
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025378383243785654,
+      "loss": 0.0155,
+      "step": 6488
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002537698353202956,
+      "loss": 0.0153,
+      "step": 6489
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025375583646958255,
+      "loss": 0.0314,
+      "step": 6490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002537418358859512,
+      "loss": 0.0153,
+      "step": 6491
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025372783356963533,
+      "loss": 0.0046,
+      "step": 6492
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025371382952086885,
+      "loss": 0.0081,
+      "step": 6493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536998237398856,
+      "loss": 0.0063,
+      "step": 6494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536858162269196,
+      "loss": 0.0226,
+      "step": 6495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025367180698220476,
+      "loss": 0.01,
+      "step": 6496
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000253657796005975,
+      "loss": 0.0021,
+      "step": 6497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536437832984644,
+      "loss": 0.0166,
+      "step": 6498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000253629768859907,
+      "loss": 0.0126,
+      "step": 6499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536157526905368,
+      "loss": 0.024,
+      "step": 6500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025360173479058797,
+      "loss": 0.0051,
+      "step": 6501
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002535877151602946,
+      "loss": 0.0208,
+      "step": 6502
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002535736937998909,
+      "loss": 0.0043,
+      "step": 6503
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025355967070961094,
+      "loss": 0.0101,
+      "step": 6504
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000253545645889689,
+      "loss": 0.0161,
+      "step": 6505
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002535316193403593,
+      "loss": 0.0103,
+      "step": 6506
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002535175910618562,
+      "loss": 0.0034,
+      "step": 6507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002535035610544139,
+      "loss": 0.0201,
+      "step": 6508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534895293182667,
+      "loss": 0.0143,
+      "step": 6509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025347549585364907,
+      "loss": 0.0131,
+      "step": 6510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534614606607953,
+      "loss": 0.0246,
+      "step": 6511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025344742373993985,
+      "loss": 0.0207,
+      "step": 6512
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534333850913172,
+      "loss": 0.0034,
+      "step": 6513
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534193447151617,
+      "loss": 0.0062,
+      "step": 6514
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000253405302611708,
+      "loss": 0.0133,
+      "step": 6515
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533912587811905,
+      "loss": 0.0114,
+      "step": 6516
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533772132238438,
+      "loss": 0.0138,
+      "step": 6517
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533631659399025,
+      "loss": 0.0185,
+      "step": 6518
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025334911692960125,
+      "loss": 0.0209,
+      "step": 6519
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533350661931747,
+      "loss": 0.0174,
+      "step": 6520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025332101373085736,
+      "loss": 0.0194,
+      "step": 6521
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533069595428842,
+      "loss": 0.0193,
+      "step": 6522
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532929036294897,
+      "loss": 0.0252,
+      "step": 6523
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532788459909088,
+      "loss": 0.0088,
+      "step": 6524
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532647866273762,
+      "loss": 0.0125,
+      "step": 6525
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532507255391267,
+      "loss": 0.016,
+      "step": 6526
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025323666272639525,
+      "loss": 0.0224,
+      "step": 6527
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532225981894166,
+      "loss": 0.0106,
+      "step": 6528
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532085319284258,
+      "loss": 0.0178,
+      "step": 6529
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025319446394365757,
+      "loss": 0.0083,
+      "step": 6530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002531803942353471,
+      "loss": 0.0093,
+      "step": 6531
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025316632280372926,
+      "loss": 0.0106,
+      "step": 6532
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025315224964903905,
+      "loss": 0.0123,
+      "step": 6533
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002531381747715116,
+      "loss": 0.0116,
+      "step": 6534
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025312409817138196,
+      "loss": 0.0157,
+      "step": 6535
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025311001984888514,
+      "loss": 0.0045,
+      "step": 6536
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025309593980425645,
+      "loss": 0.0114,
+      "step": 6537
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025308185803773087,
+      "loss": 0.0064,
+      "step": 6538
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025306777454954375,
+      "loss": 0.0137,
+      "step": 6539
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025305368933993024,
+      "loss": 0.0041,
+      "step": 6540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025303960240912555,
+      "loss": 0.0331,
+      "step": 6541
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000253025513757365,
+      "loss": 0.0016,
+      "step": 6542
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002530114233848839,
+      "loss": 0.0375,
+      "step": 6543
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529973312919176,
+      "loss": 0.0052,
+      "step": 6544
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529832374787014,
+      "loss": 0.0184,
+      "step": 6545
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529691419454708,
+      "loss": 0.0201,
+      "step": 6546
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529550446924611,
+      "loss": 0.035,
+      "step": 6547
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529409457199079,
+      "loss": 0.0154,
+      "step": 6548
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529268450280465,
+      "loss": 0.0273,
+      "step": 6549
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025291274261711255,
+      "loss": 0.0177,
+      "step": 6550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025289863848734155,
+      "loss": 0.0185,
+      "step": 6551
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000252884532638969,
+      "loss": 0.0026,
+      "step": 6552
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002528704250722306,
+      "loss": 0.0259,
+      "step": 6553
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002528563157873619,
+      "loss": 0.0096,
+      "step": 6554
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025284220478459854,
+      "loss": 0.0128,
+      "step": 6555
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025282809206417624,
+      "loss": 0.0064,
+      "step": 6556
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002528139776263307,
+      "loss": 0.0146,
+      "step": 6557
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025279986147129773,
+      "loss": 0.0067,
+      "step": 6558
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025278574359931293,
+      "loss": 0.0159,
+      "step": 6559
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002527716240106122,
+      "loss": 0.0146,
+      "step": 6560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002527575027054315,
+      "loss": 0.0148,
+      "step": 6561
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025274337968400634,
+      "loss": 0.0066,
+      "step": 6562
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002527292549465729,
+      "loss": 0.0167,
+      "step": 6563
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025271512849336697,
+      "loss": 0.016,
+      "step": 6564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002527010003246245,
+      "loss": 0.0076,
+      "step": 6565
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025268687044058144,
+      "loss": 0.0069,
+      "step": 6566
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025267273884147385,
+      "loss": 0.0093,
+      "step": 6567
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526586055275377,
+      "loss": 0.0066,
+      "step": 6568
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025264447049900906,
+      "loss": 0.0178,
+      "step": 6569
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000252630333756124,
+      "loss": 0.0053,
+      "step": 6570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526161952991186,
+      "loss": 0.037,
+      "step": 6571
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526020551282291,
+      "loss": 0.0174,
+      "step": 6572
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002525879132436916,
+      "loss": 0.0411,
+      "step": 6573
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025257376964574223,
+      "loss": 0.0102,
+      "step": 6574
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025255962433461734,
+      "loss": 0.0071,
+      "step": 6575
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025254547731055313,
+      "loss": 0.0171,
+      "step": 6576
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025253132857378587,
+      "loss": 0.0189,
+      "step": 6577
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002525171781245519,
+      "loss": 0.0109,
+      "step": 6578
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025250302596308753,
+      "loss": 0.0133,
+      "step": 6579
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002524888720896291,
+      "loss": 0.0079,
+      "step": 6580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002524747165044131,
+      "loss": 0.0161,
+      "step": 6581
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025246055920767586,
+      "loss": 0.0176,
+      "step": 6582
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025244640019965394,
+      "loss": 0.0086,
+      "step": 6583
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025243223948058366,
+      "loss": 0.0112,
+      "step": 6584
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025241807705070167,
+      "loss": 0.0054,
+      "step": 6585
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025240391291024447,
+      "loss": 0.0242,
+      "step": 6586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025238974705944864,
+      "loss": 0.0069,
+      "step": 6587
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025237557949855076,
+      "loss": 0.0116,
+      "step": 6588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002523614102277874,
+      "loss": 0.0173,
+      "step": 6589
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002523472392473953,
+      "loss": 0.0182,
+      "step": 6590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025233306655761113,
+      "loss": 0.0133,
+      "step": 6591
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025231889215867156,
+      "loss": 0.0185,
+      "step": 6592
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025230471605081333,
+      "loss": 0.016,
+      "step": 6593
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025229053823427324,
+      "loss": 0.0014,
+      "step": 6594
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522763587092881,
+      "loss": 0.0177,
+      "step": 6595
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522621774760947,
+      "loss": 0.0019,
+      "step": 6596
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522479945349299,
+      "loss": 0.0113,
+      "step": 6597
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025223380988603056,
+      "loss": 0.0037,
+      "step": 6598
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025221962352963367,
+      "loss": 0.0199,
+      "step": 6599
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522054354659761,
+      "loss": 0.0148,
+      "step": 6600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002521912456952948,
+      "loss": 0.0069,
+      "step": 6601
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002521770542178268,
+      "loss": 0.0102,
+      "step": 6602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025216286103380914,
+      "loss": 0.0244,
+      "step": 6603
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025214866614347886,
+      "loss": 0.0207,
+      "step": 6604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025213446954707304,
+      "loss": 0.0035,
+      "step": 6605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025212027124482873,
+      "loss": 0.0167,
+      "step": 6606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025210607123698315,
+      "loss": 0.0104,
+      "step": 6607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520918695237735,
+      "loss": 0.0083,
+      "step": 6608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025207766610543683,
+      "loss": 0.0158,
+      "step": 6609
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025206346098221054,
+      "loss": 0.0149,
+      "step": 6610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520492541543317,
+      "loss": 0.0117,
+      "step": 6611
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025203504562203773,
+      "loss": 0.0486,
+      "step": 6612
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025202083538556594,
+      "loss": 0.0079,
+      "step": 6613
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025200662344515356,
+      "loss": 0.0269,
+      "step": 6614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025199240980103803,
+      "loss": 0.0091,
+      "step": 6615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025197819445345673,
+      "loss": 0.0118,
+      "step": 6616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025196397740264706,
+      "loss": 0.0262,
+      "step": 6617
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025194975864884655,
+      "loss": 0.016,
+      "step": 6618
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002519355381922926,
+      "loss": 0.0035,
+      "step": 6619
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025192131603322276,
+      "loss": 0.0126,
+      "step": 6620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002519070921718745,
+      "loss": 0.0016,
+      "step": 6621
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518928666084855,
+      "loss": 0.0074,
+      "step": 6622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518786393432933,
+      "loss": 0.0051,
+      "step": 6623
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025186441037653546,
+      "loss": 0.0054,
+      "step": 6624
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518501797084497,
+      "loss": 0.0237,
+      "step": 6625
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025183594733927365,
+      "loss": 0.0234,
+      "step": 6626
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025182171326924515,
+      "loss": 0.0144,
+      "step": 6627
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518074774986018,
+      "loss": 0.0129,
+      "step": 6628
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025179324002758135,
+      "loss": 0.0386,
+      "step": 6629
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517790008564217,
+      "loss": 0.0077,
+      "step": 6630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517647599853606,
+      "loss": 0.0124,
+      "step": 6631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025175051741463584,
+      "loss": 0.0086,
+      "step": 6632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025173627314448545,
+      "loss": 0.0406,
+      "step": 6633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025172202717514724,
+      "loss": 0.0096,
+      "step": 6634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517077795068591,
+      "loss": 0.023,
+      "step": 6635
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025169353013985914,
+      "loss": 0.0268,
+      "step": 6636
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516792790743853,
+      "loss": 0.0045,
+      "step": 6637
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025166502631067546,
+      "loss": 0.0354,
+      "step": 6638
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025165077184896784,
+      "loss": 0.0139,
+      "step": 6639
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516365156895004,
+      "loss": 0.0121,
+      "step": 6640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516222578325113,
+      "loss": 0.0115,
+      "step": 6641
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025160799827823876,
+      "loss": 0.0105,
+      "step": 6642
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025159373702692076,
+      "loss": 0.0108,
+      "step": 6643
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515794740787956,
+      "loss": 0.0156,
+      "step": 6644
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515652094341015,
+      "loss": 0.0065,
+      "step": 6645
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515509430930767,
+      "loss": 0.0041,
+      "step": 6646
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025153667505595944,
+      "loss": 0.0318,
+      "step": 6647
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025152240532298796,
+      "loss": 0.0172,
+      "step": 6648
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515081338944008,
+      "loss": 0.0075,
+      "step": 6649
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002514938607704361,
+      "loss": 0.0214,
+      "step": 6650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025147958595133236,
+      "loss": 0.0086,
+      "step": 6651
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025146530943732806,
+      "loss": 0.0099,
+      "step": 6652
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025145103122866145,
+      "loss": 0.0042,
+      "step": 6653
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002514367513255712,
+      "loss": 0.0085,
+      "step": 6654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025142246972829565,
+      "loss": 0.0124,
+      "step": 6655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002514081864370735,
+      "loss": 0.0234,
+      "step": 6656
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025139390145214317,
+      "loss": 0.0356,
+      "step": 6657
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002513796147737433,
+      "loss": 0.0119,
+      "step": 6658
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025136532640211245,
+      "loss": 0.0067,
+      "step": 6659
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025135103633748936,
+      "loss": 0.0179,
+      "step": 6660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025133674458011263,
+      "loss": 0.0047,
+      "step": 6661
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025132245113022097,
+      "loss": 0.0178,
+      "step": 6662
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025130815598805315,
+      "loss": 0.0114,
+      "step": 6663
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512938591538478,
+      "loss": 0.0104,
+      "step": 6664
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025127956062784393,
+      "loss": 0.0177,
+      "step": 6665
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512652604102802,
+      "loss": 0.0221,
+      "step": 6666
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025125095850139534,
+      "loss": 0.0166,
+      "step": 6667
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025123665490142845,
+      "loss": 0.0047,
+      "step": 6668
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512223496106183,
+      "loss": 0.0441,
+      "step": 6669
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512080426292039,
+      "loss": 0.0094,
+      "step": 6670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025119373395742406,
+      "loss": 0.0142,
+      "step": 6671
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511794235955179,
+      "loss": 0.0092,
+      "step": 6672
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511651115437243,
+      "loss": 0.0063,
+      "step": 6673
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025115079780228247,
+      "loss": 0.0117,
+      "step": 6674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511364823714313,
+      "loss": 0.0156,
+      "step": 6675
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025112216525141005,
+      "loss": 0.032,
+      "step": 6676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511078464424577,
+      "loss": 0.0439,
+      "step": 6677
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510935259448135,
+      "loss": 0.0255,
+      "step": 6678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510792037587166,
+      "loss": 0.0162,
+      "step": 6679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510648798844061,
+      "loss": 0.0083,
+      "step": 6680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510505543221214,
+      "loss": 0.0101,
+      "step": 6681
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510362270721017,
+      "loss": 0.0201,
+      "step": 6682
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025102189813458627,
+      "loss": 0.0164,
+      "step": 6683
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025100756750981446,
+      "loss": 0.0127,
+      "step": 6684
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002509932351980256,
+      "loss": 0.035,
+      "step": 6685
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002509789011994591,
+      "loss": 0.0121,
+      "step": 6686
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025096456551435426,
+      "loss": 0.0207,
+      "step": 6687
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002509502281429507,
+      "loss": 0.0205,
+      "step": 6688
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002509358890854877,
+      "loss": 0.0146,
+      "step": 6689
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025092154834220485,
+      "loss": 0.0083,
+      "step": 6690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025090720591334164,
+      "loss": 0.0038,
+      "step": 6691
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002508928617991376,
+      "loss": 0.0031,
+      "step": 6692
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025087851599983233,
+      "loss": 0.0081,
+      "step": 6693
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025086416851566545,
+      "loss": 0.025,
+      "step": 6694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002508498193468765,
+      "loss": 0.0322,
+      "step": 6695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025083546849370524,
+      "loss": 0.0142,
+      "step": 6696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002508211159563913,
+      "loss": 0.0085,
+      "step": 6697
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025080676173517446,
+      "loss": 0.0095,
+      "step": 6698
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025079240583029434,
+      "loss": 0.0084,
+      "step": 6699
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507780482419908,
+      "loss": 0.0073,
+      "step": 6700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507636889705036,
+      "loss": 0.016,
+      "step": 6701
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507493280160727,
+      "loss": 0.0252,
+      "step": 6702
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507349653789378,
+      "loss": 0.0382,
+      "step": 6703
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002507206010593387,
+      "loss": 0.0177,
+      "step": 6704
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025070623505751556,
+      "loss": 0.0323,
+      "step": 6705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025069186737370817,
+      "loss": 0.0049,
+      "step": 6706
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025067749800815656,
+      "loss": 0.0134,
+      "step": 6707
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025066312696110063,
+      "loss": 0.006,
+      "step": 6708
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002506487542327805,
+      "loss": 0.0382,
+      "step": 6709
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025063437982343617,
+      "loss": 0.0064,
+      "step": 6710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002506200037333077,
+      "loss": 0.0119,
+      "step": 6711
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025060562596263534,
+      "loss": 0.0047,
+      "step": 6712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025059124651165905,
+      "loss": 0.0114,
+      "step": 6713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002505768653806191,
+      "loss": 0.0142,
+      "step": 6714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002505624825697556,
+      "loss": 0.0162,
+      "step": 6715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025054809807930885,
+      "loss": 0.0125,
+      "step": 6716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025053371190951907,
+      "loss": 0.0105,
+      "step": 6717
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025051932406062647,
+      "loss": 0.0203,
+      "step": 6718
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025050493453287147,
+      "loss": 0.0254,
+      "step": 6719
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504905433264944,
+      "loss": 0.0096,
+      "step": 6720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504761504417354,
+      "loss": 0.0087,
+      "step": 6721
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025046175587883516,
+      "loss": 0.022,
+      "step": 6722
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504473596380339,
+      "loss": 0.018,
+      "step": 6723
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504329617195722,
+      "loss": 0.0108,
+      "step": 6724
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504185621236904,
+      "loss": 0.0223,
+      "step": 6725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025040416085062906,
+      "loss": 0.0237,
+      "step": 6726
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002503897579006287,
+      "loss": 0.0098,
+      "step": 6727
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002503753532739299,
+      "loss": 0.0214,
+      "step": 6728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025036094697077326,
+      "loss": 0.0132,
+      "step": 6729
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025034653899139934,
+      "loss": 0.0225,
+      "step": 6730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002503321293360488,
+      "loss": 0.0054,
+      "step": 6731
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025031771800496224,
+      "loss": 0.0063,
+      "step": 6732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002503033049983805,
+      "loss": 0.0157,
+      "step": 6733
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502888903165442,
+      "loss": 0.0057,
+      "step": 6734
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502744739596942,
+      "loss": 0.0212,
+      "step": 6735
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502600559280711,
+      "loss": 0.0231,
+      "step": 6736
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502456362219159,
+      "loss": 0.0115,
+      "step": 6737
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502312148414693,
+      "loss": 0.0084,
+      "step": 6738
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502167917869722,
+      "loss": 0.0087,
+      "step": 6739
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025020236705866557,
+      "loss": 0.011,
+      "step": 6740
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025018794065679017,
+      "loss": 0.0082,
+      "step": 6741
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025017351258158706,
+      "loss": 0.023,
+      "step": 6742
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501590828332972,
+      "loss": 0.0021,
+      "step": 6743
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501446514121617,
+      "loss": 0.0121,
+      "step": 6744
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501302183184214,
+      "loss": 0.0013,
+      "step": 6745
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501157835523174,
+      "loss": 0.0175,
+      "step": 6746
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501013471140909,
+      "loss": 0.0323,
+      "step": 6747
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025008690900398293,
+      "loss": 0.0222,
+      "step": 6748
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002500724692222347,
+      "loss": 0.0146,
+      "step": 6749
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025005802776908727,
+      "loss": 0.0186,
+      "step": 6750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002500435846447819,
+      "loss": 0.0185,
+      "step": 6751
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025002913984955986,
+      "loss": 0.0214,
+      "step": 6752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00025001469338366234,
+      "loss": 0.0064,
+      "step": 6753
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00025000024524733064,
+      "loss": 0.0107,
+      "step": 6754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024998579544080614,
+      "loss": 0.0059,
+      "step": 6755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024997134396433004,
+      "loss": 0.0068,
+      "step": 6756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499568908181439,
+      "loss": 0.0057,
+      "step": 6757
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499424360024889,
+      "loss": 0.0132,
+      "step": 6758
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024992797951760656,
+      "loss": 0.0111,
+      "step": 6759
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024991352136373836,
+      "loss": 0.0135,
+      "step": 6760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024989906154112575,
+      "loss": 0.0054,
+      "step": 6761
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498846000500103,
+      "loss": 0.0084,
+      "step": 6762
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498701368906334,
+      "loss": 0.011,
+      "step": 6763
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498556720632367,
+      "loss": 0.0364,
+      "step": 6764
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498412055680618,
+      "loss": 0.0234,
+      "step": 6765
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024982673740535035,
+      "loss": 0.0091,
+      "step": 6766
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024981226757534396,
+      "loss": 0.0186,
+      "step": 6767
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002497977960782842,
+      "loss": 0.0033,
+      "step": 6768
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024978332291441293,
+      "loss": 0.0203,
+      "step": 6769
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024976884808397174,
+      "loss": 0.0065,
+      "step": 6770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024975437158720256,
+      "loss": 0.0116,
+      "step": 6771
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000249739893424347,
+      "loss": 0.0082,
+      "step": 6772
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000249725413595647,
+      "loss": 0.0284,
+      "step": 6773
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002497109321013443,
+      "loss": 0.025,
+      "step": 6774
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024969644894168086,
+      "loss": 0.047,
+      "step": 6775
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024968196411689854,
+      "loss": 0.0089,
+      "step": 6776
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002496674776272392,
+      "loss": 0.014,
+      "step": 6777
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002496529894729449,
+      "loss": 0.0182,
+      "step": 6778
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024963849965425753,
+      "loss": 0.0211,
+      "step": 6779
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024962400817141916,
+      "loss": 0.014,
+      "step": 6780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024960951502467176,
+      "loss": 0.0204,
+      "step": 6781
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495950202142575,
+      "loss": 0.0094,
+      "step": 6782
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024958052374041834,
+      "loss": 0.0088,
+      "step": 6783
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024956602560339647,
+      "loss": 0.0165,
+      "step": 6784
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495515258034341,
+      "loss": 0.0262,
+      "step": 6785
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024953702434077325,
+      "loss": 0.0112,
+      "step": 6786
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495225212156562,
+      "loss": 0.0333,
+      "step": 6787
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024950801642832524,
+      "loss": 0.0115,
+      "step": 6788
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494935099790225,
+      "loss": 0.039,
+      "step": 6789
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024947900186799045,
+      "loss": 0.0092,
+      "step": 6790
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494644920954712,
+      "loss": 0.0091,
+      "step": 6791
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024944998066170717,
+      "loss": 0.0101,
+      "step": 6792
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494354675669408,
+      "loss": 0.0051,
+      "step": 6793
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494209528114144,
+      "loss": 0.0173,
+      "step": 6794
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494064363953704,
+      "loss": 0.0029,
+      "step": 6795
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493919183190513,
+      "loss": 0.0106,
+      "step": 6796
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024937739858269946,
+      "loss": 0.0106,
+      "step": 6797
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024936287718655754,
+      "loss": 0.0018,
+      "step": 6798
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024934835413086806,
+      "loss": 0.0104,
+      "step": 6799
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493338294158735,
+      "loss": 0.0359,
+      "step": 6800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493193030418165,
+      "loss": 0.0268,
+      "step": 6801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024930477500893956,
+      "loss": 0.018,
+      "step": 6802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492902453174855,
+      "loss": 0.0187,
+      "step": 6803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024927571396769695,
+      "loss": 0.0247,
+      "step": 6804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024926118095981655,
+      "loss": 0.0153,
+      "step": 6805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024924664629408705,
+      "loss": 0.0112,
+      "step": 6806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492321099707512,
+      "loss": 0.009,
+      "step": 6807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492175719900518,
+      "loss": 0.0833,
+      "step": 6808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492030323522317,
+      "loss": 0.0077,
+      "step": 6809
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002491884910575337,
+      "loss": 0.0069,
+      "step": 6810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024917394810620064,
+      "loss": 0.0236,
+      "step": 6811
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024915940349847545,
+      "loss": 0.02,
+      "step": 6812
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024914485723460104,
+      "loss": 0.003,
+      "step": 6813
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002491303093148204,
+      "loss": 0.0135,
+      "step": 6814
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024911575973937645,
+      "loss": 0.0099,
+      "step": 6815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024910120850851216,
+      "loss": 0.0026,
+      "step": 6816
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024908665562247074,
+      "loss": 0.0291,
+      "step": 6817
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024907210108149504,
+      "loss": 0.0084,
+      "step": 6818
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024905754488582825,
+      "loss": 0.0175,
+      "step": 6819
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002490429870357135,
+      "loss": 0.0116,
+      "step": 6820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002490284275313939,
+      "loss": 0.0147,
+      "step": 6821
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024901386637311265,
+      "loss": 0.012,
+      "step": 6822
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002489993035611129,
+      "loss": 0.0086,
+      "step": 6823
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024898473909563794,
+      "loss": 0.0078,
+      "step": 6824
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024897017297693095,
+      "loss": 0.021,
+      "step": 6825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024895560520523524,
+      "loss": 0.0172,
+      "step": 6826
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002489410357807942,
+      "loss": 0.007,
+      "step": 6827
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024892646470385104,
+      "loss": 0.0111,
+      "step": 6828
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024891189197464924,
+      "loss": 0.0294,
+      "step": 6829
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024889731759343206,
+      "loss": 0.0123,
+      "step": 6830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024888274156044306,
+      "loss": 0.0097,
+      "step": 6831
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002488681638759256,
+      "loss": 0.0044,
+      "step": 6832
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002488535845401231,
+      "loss": 0.0099,
+      "step": 6833
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002488390035532793,
+      "loss": 0.0029,
+      "step": 6834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024882442091563744,
+      "loss": 0.0353,
+      "step": 6835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024880983662744127,
+      "loss": 0.0082,
+      "step": 6836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024879525068893424,
+      "loss": 0.0185,
+      "step": 6837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002487806631003601,
+      "loss": 0.0246,
+      "step": 6838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024876607386196243,
+      "loss": 0.005,
+      "step": 6839
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002487514829739848,
+      "loss": 0.0132,
+      "step": 6840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002487368904366711,
+      "loss": 0.0244,
+      "step": 6841
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024872229625026493,
+      "loss": 0.045,
+      "step": 6842
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024870770041501004,
+      "loss": 0.033,
+      "step": 6843
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024869310293115023,
+      "loss": 0.0194,
+      "step": 6844
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024867850379892934,
+      "loss": 0.0113,
+      "step": 6845
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486639030185911,
+      "loss": 0.0115,
+      "step": 6846
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486493005903795,
+      "loss": 0.0045,
+      "step": 6847
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024863469651453834,
+      "loss": 0.0024,
+      "step": 6848
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024862009079131155,
+      "loss": 0.0037,
+      "step": 6849
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486054834209431,
+      "loss": 0.0139,
+      "step": 6850
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024859087440367694,
+      "loss": 0.0315,
+      "step": 6851
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002485762637397571,
+      "loss": 0.0287,
+      "step": 6852
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002485616514294275,
+      "loss": 0.0148,
+      "step": 6853
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024854703747293237,
+      "loss": 0.0068,
+      "step": 6854
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024853242187051567,
+      "loss": 0.0118,
+      "step": 6855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002485178046224215,
+      "loss": 0.013,
+      "step": 6856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024850318572889403,
+      "loss": 0.0034,
+      "step": 6857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484885651901774,
+      "loss": 0.0108,
+      "step": 6858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484739430065159,
+      "loss": 0.0125,
+      "step": 6859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484593191781536,
+      "loss": 0.016,
+      "step": 6860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484446937053349,
+      "loss": 0.0145,
+      "step": 6861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484300665883039,
+      "loss": 0.0253,
+      "step": 6862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000248415437827305,
+      "loss": 0.0165,
+      "step": 6863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024840080742258254,
+      "loss": 0.0058,
+      "step": 6864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024838617537438083,
+      "loss": 0.0418,
+      "step": 6865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483715416829443,
+      "loss": 0.01,
+      "step": 6866
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483569063485173,
+      "loss": 0.0099,
+      "step": 6867
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483422693713443,
+      "loss": 0.0125,
+      "step": 6868
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483276307516698,
+      "loss": 0.01,
+      "step": 6869
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024831299048973823,
+      "loss": 0.0119,
+      "step": 6870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024829834858579417,
+      "loss": 0.014,
+      "step": 6871
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024828370504008216,
+      "loss": 0.0212,
+      "step": 6872
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024826905985284676,
+      "loss": 0.011,
+      "step": 6873
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002482544130243325,
+      "loss": 0.017,
+      "step": 6874
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002482397645547841,
+      "loss": 0.008,
+      "step": 6875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002482251144444462,
+      "loss": 0.0273,
+      "step": 6876
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024821046269356343,
+      "loss": 0.0101,
+      "step": 6877
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481958093023806,
+      "loss": 0.0076,
+      "step": 6878
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024818115427114234,
+      "loss": 0.0283,
+      "step": 6879
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024816649760009353,
+      "loss": 0.0053,
+      "step": 6880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481518392894789,
+      "loss": 0.0239,
+      "step": 6881
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024813717933954327,
+      "loss": 0.0075,
+      "step": 6882
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481225177505315,
+      "loss": 0.0147,
+      "step": 6883
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024810785452268846,
+      "loss": 0.0097,
+      "step": 6884
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002480931896562591,
+      "loss": 0.0096,
+      "step": 6885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002480785231514882,
+      "loss": 0.0099,
+      "step": 6886
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024806385500862095,
+      "loss": 0.0099,
+      "step": 6887
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024804918522790216,
+      "loss": 0.0192,
+      "step": 6888
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024803451380957687,
+      "loss": 0.0115,
+      "step": 6889
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024801984075389016,
+      "loss": 0.0239,
+      "step": 6890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002480051660610871,
+      "loss": 0.0117,
+      "step": 6891
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479904897314128,
+      "loss": 0.0096,
+      "step": 6892
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024797581176511227,
+      "loss": 0.0277,
+      "step": 6893
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479611321624308,
+      "loss": 0.0231,
+      "step": 6894
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024794645092361347,
+      "loss": 0.0096,
+      "step": 6895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024793176804890557,
+      "loss": 0.0034,
+      "step": 6896
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479170835385522,
+      "loss": 0.0182,
+      "step": 6897
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479023973927988,
+      "loss": 0.0124,
+      "step": 6898
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024788770961189054,
+      "loss": 0.0268,
+      "step": 6899
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024787302019607275,
+      "loss": 0.0103,
+      "step": 6900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024785832914559077,
+      "loss": 0.0017,
+      "step": 6901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024784363646068996,
+      "loss": 0.0257,
+      "step": 6902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002478289421416157,
+      "loss": 0.0159,
+      "step": 6903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002478142461886135,
+      "loss": 0.0181,
+      "step": 6904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002477995486019287,
+      "loss": 0.0048,
+      "step": 6905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024778484938180683,
+      "loss": 0.0146,
+      "step": 6906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024777014852849345,
+      "loss": 0.0088,
+      "step": 6907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024775544604223397,
+      "loss": 0.0229,
+      "step": 6908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002477407419232741,
+      "loss": 0.0129,
+      "step": 6909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024772603617185927,
+      "loss": 0.0043,
+      "step": 6910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002477113287882352,
+      "loss": 0.004,
+      "step": 6911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476966197726475,
+      "loss": 0.027,
+      "step": 6912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024768190912534177,
+      "loss": 0.0077,
+      "step": 6913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024766719684656376,
+      "loss": 0.0089,
+      "step": 6914
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476524829365593,
+      "loss": 0.0095,
+      "step": 6915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000247637767395574,
+      "loss": 0.0299,
+      "step": 6916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024762305022385365,
+      "loss": 0.0417,
+      "step": 6917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024760833142164416,
+      "loss": 0.0265,
+      "step": 6918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475936109891912,
+      "loss": 0.0259,
+      "step": 6919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024757888892674074,
+      "loss": 0.0077,
+      "step": 6920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475641652345387,
+      "loss": 0.0238,
+      "step": 6921
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024754943991283084,
+      "loss": 0.0118,
+      "step": 6922
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024753471296186326,
+      "loss": 0.0006,
+      "step": 6923
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475199843818819,
+      "loss": 0.0091,
+      "step": 6924
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024750525417313265,
+      "loss": 0.0117,
+      "step": 6925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024749052233586163,
+      "loss": 0.0066,
+      "step": 6926
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474757888703148,
+      "loss": 0.0037,
+      "step": 6927
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024746105377673836,
+      "loss": 0.02,
+      "step": 6928
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474463170553784,
+      "loss": 0.005,
+      "step": 6929
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474315787064809,
+      "loss": 0.0074,
+      "step": 6930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024741683873029217,
+      "loss": 0.0264,
+      "step": 6931
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474020971270583,
+      "loss": 0.0125,
+      "step": 6932
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473873538970256,
+      "loss": 0.0082,
+      "step": 6933
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024737260904044025,
+      "loss": 0.0218,
+      "step": 6934
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473578625575485,
+      "loss": 0.02,
+      "step": 6935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473431144485967,
+      "loss": 0.0188,
+      "step": 6936
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473283647138312,
+      "loss": 0.0136,
+      "step": 6937
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473136133534982,
+      "loss": 0.0142,
+      "step": 6938
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024729886036784415,
+      "loss": 0.0052,
+      "step": 6939
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024728410575711553,
+      "loss": 0.0128,
+      "step": 6940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024726934952155866,
+      "loss": 0.0142,
+      "step": 6941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472545916614201,
+      "loss": 0.0399,
+      "step": 6942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024723983217694623,
+      "loss": 0.0113,
+      "step": 6943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472250710683836,
+      "loss": 0.0082,
+      "step": 6944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472103083359788,
+      "loss": 0.0093,
+      "step": 6945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024719554397997827,
+      "loss": 0.0202,
+      "step": 6946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024718077800062877,
+      "loss": 0.0136,
+      "step": 6947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024716601039817684,
+      "loss": 0.02,
+      "step": 6948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002471512411728691,
+      "loss": 0.0179,
+      "step": 6949
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024713647032495223,
+      "loss": 0.0165,
+      "step": 6950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002471216978546729,
+      "loss": 0.012,
+      "step": 6951
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024710692376227796,
+      "loss": 0.0101,
+      "step": 6952
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024709214804801414,
+      "loss": 0.0477,
+      "step": 6953
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002470773707121281,
+      "loss": 0.0037,
+      "step": 6954
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002470625917548667,
+      "loss": 0.0246,
+      "step": 6955
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024704781117647687,
+      "loss": 0.0142,
+      "step": 6956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024703302897720536,
+      "loss": 0.0284,
+      "step": 6957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002470182451572991,
+      "loss": 0.0084,
+      "step": 6958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000247003459717005,
+      "loss": 0.0142,
+      "step": 6959
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024698867265657,
+      "loss": 0.0058,
+      "step": 6960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024697388397624116,
+      "loss": 0.0091,
+      "step": 6961
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024695909367626536,
+      "loss": 0.0168,
+      "step": 6962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469443017568897,
+      "loss": 0.0095,
+      "step": 6963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469295082183612,
+      "loss": 0.0301,
+      "step": 6964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469147130609269,
+      "loss": 0.0123,
+      "step": 6965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000246899916284834,
+      "loss": 0.0163,
+      "step": 6966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024688511789032957,
+      "loss": 0.0143,
+      "step": 6967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024687031787766077,
+      "loss": 0.0085,
+      "step": 6968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002468555162470748,
+      "loss": 0.0162,
+      "step": 6969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024684071299881896,
+      "loss": 0.0175,
+      "step": 6970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024682590813314035,
+      "loss": 0.0207,
+      "step": 6971
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002468111016502863,
+      "loss": 0.0236,
+      "step": 6972
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024679629355050414,
+      "loss": 0.0097,
+      "step": 6973
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002467814838340411,
+      "loss": 0.0415,
+      "step": 6974
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024676667250114466,
+      "loss": 0.0125,
+      "step": 6975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024675185955206204,
+      "loss": 0.0031,
+      "step": 6976
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002467370449870408,
+      "loss": 0.0038,
+      "step": 6977
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002467222288063283,
+      "loss": 0.0071,
+      "step": 6978
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000246707411010172,
+      "loss": 0.0181,
+      "step": 6979
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002466925915988194,
+      "loss": 0.0175,
+      "step": 6980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000246677770572518,
+      "loss": 0.0166,
+      "step": 6981
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002466629479315153,
+      "loss": 0.007,
+      "step": 6982
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024664812367605895,
+      "loss": 0.0052,
+      "step": 6983
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024663329780639647,
+      "loss": 0.0111,
+      "step": 6984
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024661847032277556,
+      "loss": 0.0226,
+      "step": 6985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024660364122544375,
+      "loss": 0.0334,
+      "step": 6986
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024658881051464884,
+      "loss": 0.0316,
+      "step": 6987
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002465739781906384,
+      "loss": 0.0127,
+      "step": 6988
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024655914425366027,
+      "loss": 0.0118,
+      "step": 6989
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024654430870396215,
+      "loss": 0.0154,
+      "step": 6990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024652947154179183,
+      "loss": 0.0101,
+      "step": 6991
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024651463276739716,
+      "loss": 0.0031,
+      "step": 6992
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024649979238102593,
+      "loss": 0.0255,
+      "step": 6993
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024648495038292594,
+      "loss": 0.0144,
+      "step": 6994
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024647010677334526,
+      "loss": 0.0231,
+      "step": 6995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002464552615525316,
+      "loss": 0.0143,
+      "step": 6996
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000246440414720733,
+      "loss": 0.0107,
+      "step": 6997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002464255662781975,
+      "loss": 0.0332,
+      "step": 6998
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246410716225173,
+      "loss": 0.0066,
+      "step": 6999
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002463958645619075,
+      "loss": 0.006,
+      "step": 7000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024638101128864916,
+      "loss": 0.0061,
+      "step": 7001
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246366156405646,
+      "loss": 0.0062,
+      "step": 7002
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024635129991314614,
+      "loss": 0.0189,
+      "step": 7003
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002463364418113977,
+      "loss": 0.0154,
+      "step": 7004
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002463215821006488,
+      "loss": 0.0365,
+      "step": 7005
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002463067207811476,
+      "loss": 0.0155,
+      "step": 7006
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462918578531425,
+      "loss": 0.008,
+      "step": 7007
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462769933168815,
+      "loss": 0.0109,
+      "step": 7008
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246262127172613,
+      "loss": 0.0321,
+      "step": 7009
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462472594205853,
+      "loss": 0.0104,
+      "step": 7010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462323900610467,
+      "loss": 0.0238,
+      "step": 7011
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462175190942455,
+      "loss": 0.0092,
+      "step": 7012
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462026465204301,
+      "loss": 0.0104,
+      "step": 7013
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024618777233984896,
+      "loss": 0.012,
+      "step": 7014
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461728965527504,
+      "loss": 0.0239,
+      "step": 7015
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024615801915938297,
+      "loss": 0.0106,
+      "step": 7016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461431401599951,
+      "loss": 0.0175,
+      "step": 7017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461282595548353,
+      "loss": 0.0168,
+      "step": 7018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461133773441521,
+      "loss": 0.0193,
+      "step": 7019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246098493528194,
+      "loss": 0.0267,
+      "step": 7020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002460836081072098,
+      "loss": 0.0082,
+      "step": 7021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024606872108144786,
+      "loss": 0.02,
+      "step": 7022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024605383245115704,
+      "loss": 0.0056,
+      "step": 7023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002460389422165858,
+      "loss": 0.0089,
+      "step": 7024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246024050377983,
+      "loss": 0.0017,
+      "step": 7025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024600915693559727,
+      "loss": 0.0087,
+      "step": 7026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002459942618896774,
+      "loss": 0.0171,
+      "step": 7027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024597936524047215,
+      "loss": 0.0205,
+      "step": 7028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024596446698823036,
+      "loss": 0.0017,
+      "step": 7029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002459495671332008,
+      "loss": 0.0567,
+      "step": 7030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024593466567563235,
+      "loss": 0.0107,
+      "step": 7031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002459197626157739,
+      "loss": 0.016,
+      "step": 7032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024590485795387436,
+      "loss": 0.0066,
+      "step": 7033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458899516901826,
+      "loss": 0.0477,
+      "step": 7034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024587504382494774,
+      "loss": 0.0098,
+      "step": 7035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458601343584186,
+      "loss": 0.0338,
+      "step": 7036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024584522329084427,
+      "loss": 0.0154,
+      "step": 7037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458303106224738,
+      "loss": 0.0151,
+      "step": 7038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458153963535563,
+      "loss": 0.0292,
+      "step": 7039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458004804843408,
+      "loss": 0.0089,
+      "step": 7040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002457855630150764,
+      "loss": 0.0038,
+      "step": 7041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024577064394601226,
+      "loss": 0.0133,
+      "step": 7042
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002457557232773976,
+      "loss": 0.0144,
+      "step": 7043
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002457408010094817,
+      "loss": 0.0052,
+      "step": 7044
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024572587714251365,
+      "loss": 0.0134,
+      "step": 7045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024571095167674273,
+      "loss": 0.0101,
+      "step": 7046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002456960246124182,
+      "loss": 0.0193,
+      "step": 7047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024568109594978954,
+      "loss": 0.0041,
+      "step": 7048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024566616568910595,
+      "loss": 0.0085,
+      "step": 7049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024565123383061676,
+      "loss": 0.0174,
+      "step": 7050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024563630037457144,
+      "loss": 0.0058,
+      "step": 7051
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002456213653212193,
+      "loss": 0.0164,
+      "step": 7052
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024560642867080994,
+      "loss": 0.0108,
+      "step": 7053
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024559149042359277,
+      "loss": 0.018,
+      "step": 7054
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455765505798172,
+      "loss": 0.0115,
+      "step": 7055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455616091397329,
+      "loss": 0.0123,
+      "step": 7056
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024554666610358925,
+      "loss": 0.0086,
+      "step": 7057
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024553172147163596,
+      "loss": 0.0178,
+      "step": 7058
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455167752441226,
+      "loss": 0.0107,
+      "step": 7059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455018274212988,
+      "loss": 0.037,
+      "step": 7060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454868780034142,
+      "loss": 0.0099,
+      "step": 7061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454719269907185,
+      "loss": 0.0085,
+      "step": 7062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454569743834614,
+      "loss": 0.0372,
+      "step": 7063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024544202018189263,
+      "loss": 0.0115,
+      "step": 7064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024542706438626194,
+      "loss": 0.0037,
+      "step": 7065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454121069968192,
+      "loss": 0.0018,
+      "step": 7066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024539714801381416,
+      "loss": 0.0036,
+      "step": 7067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453821874374966,
+      "loss": 0.0203,
+      "step": 7068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453672252681165,
+      "loss": 0.0338,
+      "step": 7069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453522615059238,
+      "loss": 0.0145,
+      "step": 7070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453372961511683,
+      "loss": 0.0124,
+      "step": 7071
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024532232920409993,
+      "loss": 0.0115,
+      "step": 7072
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453073606649688,
+      "loss": 0.0087,
+      "step": 7073
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452923905340248,
+      "loss": 0.0197,
+      "step": 7074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024527741881151803,
+      "loss": 0.0179,
+      "step": 7075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452624454976985,
+      "loss": 0.0168,
+      "step": 7076
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452474705928163,
+      "loss": 0.0125,
+      "step": 7077
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452324940971216,
+      "loss": 0.019,
+      "step": 7078
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452175160108644,
+      "loss": 0.0039,
+      "step": 7079
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000245202536334295,
+      "loss": 0.0073,
+      "step": 7080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024518755506766355,
+      "loss": 0.0116,
+      "step": 7081
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024517257221122024,
+      "loss": 0.0257,
+      "step": 7082
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024515758776521527,
+      "loss": 0.025,
+      "step": 7083
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000245142601729899,
+      "loss": 0.0197,
+      "step": 7084
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002451276141055217,
+      "loss": 0.0216,
+      "step": 7085
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002451126248923337,
+      "loss": 0.004,
+      "step": 7086
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024509763409058527,
+      "loss": 0.0153,
+      "step": 7087
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024508264170052684,
+      "loss": 0.0065,
+      "step": 7088
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450676477224089,
+      "loss": 0.0156,
+      "step": 7089
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450526521564817,
+      "loss": 0.0045,
+      "step": 7090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450376550029958,
+      "loss": 0.0147,
+      "step": 7091
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450226562622017,
+      "loss": 0.0121,
+      "step": 7092
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450076559343498,
+      "loss": 0.0115,
+      "step": 7093
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002449926540196908,
+      "loss": 0.0221,
+      "step": 7094
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002449776505184751,
+      "loss": 0.0046,
+      "step": 7095
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002449626454309534,
+      "loss": 0.007,
+      "step": 7096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024494763875737623,
+      "loss": 0.0199,
+      "step": 7097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024493263049799426,
+      "loss": 0.0006,
+      "step": 7098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002449176206530582,
+      "loss": 0.0174,
+      "step": 7099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002449026092228187,
+      "loss": 0.0272,
+      "step": 7100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024488759620752644,
+      "loss": 0.023,
+      "step": 7101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002448725816074323,
+      "loss": 0.0158,
+      "step": 7102
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024485756542278684,
+      "loss": 0.0069,
+      "step": 7103
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024484254765384106,
+      "loss": 0.021,
+      "step": 7104
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002448275283008457,
+      "loss": 0.0099,
+      "step": 7105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002448125073640516,
+      "loss": 0.0088,
+      "step": 7106
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447974848437097,
+      "loss": 0.0082,
+      "step": 7107
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447824607400708,
+      "loss": 0.0089,
+      "step": 7108
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447674350533859,
+      "loss": 0.0164,
+      "step": 7109
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024475240778390597,
+      "loss": 0.0089,
+      "step": 7110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000244737378931882,
+      "loss": 0.0054,
+      "step": 7111
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447223484975649,
+      "loss": 0.0289,
+      "step": 7112
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447073164812058,
+      "loss": 0.0057,
+      "step": 7113
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002446922828830558,
+      "loss": 0.0049,
+      "step": 7114
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024467724770336585,
+      "loss": 0.0078,
+      "step": 7115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024466221094238724,
+      "loss": 0.0459,
+      "step": 7116
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024464717260037097,
+      "loss": 0.0119,
+      "step": 7117
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024463213267756826,
+      "loss": 0.0121,
+      "step": 7118
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002446170911742303,
+      "loss": 0.0144,
+      "step": 7119
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024460204809060833,
+      "loss": 0.0089,
+      "step": 7120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024458700342695354,
+      "loss": 0.0051,
+      "step": 7121
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024457195718351725,
+      "loss": 0.0137,
+      "step": 7122
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024455690936055083,
+      "loss": 0.0155,
+      "step": 7123
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024454185995830544,
+      "loss": 0.0124,
+      "step": 7124
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002445268089770326,
+      "loss": 0.0148,
+      "step": 7125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002445117564169836,
+      "loss": 0.0063,
+      "step": 7126
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444967022784099,
+      "loss": 0.0251,
+      "step": 7127
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444816465615628,
+      "loss": 0.0058,
+      "step": 7128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024446658926669397,
+      "loss": 0.0079,
+      "step": 7129
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444515303940547,
+      "loss": 0.0163,
+      "step": 7130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444364699438966,
+      "loss": 0.025,
+      "step": 7131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444214079164712,
+      "loss": 0.037,
+      "step": 7132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024440634431203,
+      "loss": 0.0094,
+      "step": 7133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024439127913082474,
+      "loss": 0.0113,
+      "step": 7134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443762123731069,
+      "loss": 0.0299,
+      "step": 7135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024436114403912813,
+      "loss": 0.0133,
+      "step": 7136
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443460741291402,
+      "loss": 0.0274,
+      "step": 7137
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443310026433947,
+      "loss": 0.0109,
+      "step": 7138
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443159295821434,
+      "loss": 0.0389,
+      "step": 7139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443008549456381,
+      "loss": 0.013,
+      "step": 7140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442857787341304,
+      "loss": 0.0306,
+      "step": 7141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442707009478723,
+      "loss": 0.0165,
+      "step": 7142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442556215871155,
+      "loss": 0.028,
+      "step": 7143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024424054065211195,
+      "loss": 0.0197,
+      "step": 7144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442254581431135,
+      "loss": 0.0195,
+      "step": 7145
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024421037406037195,
+      "loss": 0.0469,
+      "step": 7146
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002441952884041394,
+      "loss": 0.0154,
+      "step": 7147
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002441802011746677,
+      "loss": 0.0187,
+      "step": 7148
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002441651123722088,
+      "loss": 0.0394,
+      "step": 7149
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024415002199701483,
+      "loss": 0.0132,
+      "step": 7150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024413493004933777,
+      "loss": 0.0114,
+      "step": 7151
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024411983652942965,
+      "loss": 0.0158,
+      "step": 7152
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024410474143754266,
+      "loss": 0.0306,
+      "step": 7153
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002440896447739288,
+      "loss": 0.0081,
+      "step": 7154
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024407454653884028,
+      "loss": 0.009,
+      "step": 7155
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024405944673252928,
+      "loss": 0.0096,
+      "step": 7156
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024404434535524796,
+      "loss": 0.0096,
+      "step": 7157
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024402924240724856,
+      "loss": 0.0126,
+      "step": 7158
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024401413788878326,
+      "loss": 0.0456,
+      "step": 7159
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024399903180010444,
+      "loss": 0.0179,
+      "step": 7160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024398392414146436,
+      "loss": 0.0278,
+      "step": 7161
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024396881491311532,
+      "loss": 0.0265,
+      "step": 7162
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002439537041153097,
+      "loss": 0.0111,
+      "step": 7163
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024393859174829985,
+      "loss": 0.0062,
+      "step": 7164
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024392347781233825,
+      "loss": 0.0125,
+      "step": 7165
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002439083623076772,
+      "loss": 0.0137,
+      "step": 7166
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002438932452345693,
+      "loss": 0.0319,
+      "step": 7167
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024387812659326695,
+      "loss": 0.0055,
+      "step": 7168
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024386300638402268,
+      "loss": 0.0145,
+      "step": 7169
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024384788460708902,
+      "loss": 0.0145,
+      "step": 7170
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024383276126271856,
+      "loss": 0.0191,
+      "step": 7171
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002438176363511638,
+      "loss": 0.0162,
+      "step": 7172
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024380250987267747,
+      "loss": 0.0017,
+      "step": 7173
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024378738182751216,
+      "loss": 0.0058,
+      "step": 7174
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024377225221592055,
+      "loss": 0.0173,
+      "step": 7175
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002437571210381553,
+      "loss": 0.0233,
+      "step": 7176
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002437419882944691,
+      "loss": 0.0167,
+      "step": 7177
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024372685398511485,
+      "loss": 0.0101,
+      "step": 7178
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024371171811034514,
+      "loss": 0.0152,
+      "step": 7179
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024369658067041285,
+      "loss": 0.0136,
+      "step": 7180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024368144166557083,
+      "loss": 0.0226,
+      "step": 7181
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024366630109607187,
+      "loss": 0.0233,
+      "step": 7182
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002436511589621689,
+      "loss": 0.0297,
+      "step": 7183
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024363601526411477,
+      "loss": 0.0346,
+      "step": 7184
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024362087000216244,
+      "loss": 0.0044,
+      "step": 7185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024360572317656487,
+      "loss": 0.0255,
+      "step": 7186
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000243590574787575,
+      "loss": 0.0229,
+      "step": 7187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024357542483544593,
+      "loss": 0.0067,
+      "step": 7188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024356027332043056,
+      "loss": 0.0365,
+      "step": 7189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024354512024278203,
+      "loss": 0.0083,
+      "step": 7190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024352996560275344,
+      "loss": 0.0189,
+      "step": 7191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024351480940059788,
+      "loss": 0.0191,
+      "step": 7192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024349965163656843,
+      "loss": 0.0109,
+      "step": 7193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024348449231091838,
+      "loss": 0.0346,
+      "step": 7194
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024346933142390078,
+      "loss": 0.0198,
+      "step": 7195
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024345416897576894,
+      "loss": 0.0057,
+      "step": 7196
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024343900496677607,
+      "loss": 0.0058,
+      "step": 7197
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024342383939717543,
+      "loss": 0.0047,
+      "step": 7198
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024340867226722032,
+      "loss": 0.0093,
+      "step": 7199
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024339350357716408,
+      "loss": 0.0073,
+      "step": 7200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024337833332726,
+      "loss": 0.005,
+      "step": 7201
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024336316151776153,
+      "loss": 0.0092,
+      "step": 7202
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024334798814892202,
+      "loss": 0.0129,
+      "step": 7203
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024333281322099488,
+      "loss": 0.0195,
+      "step": 7204
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024331763673423362,
+      "loss": 0.0133,
+      "step": 7205
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024330245868889165,
+      "loss": 0.0158,
+      "step": 7206
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024328727908522248,
+      "loss": 0.0143,
+      "step": 7207
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024327209792347968,
+      "loss": 0.012,
+      "step": 7208
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024325691520391676,
+      "loss": 0.0249,
+      "step": 7209
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024324173092678733,
+      "loss": 0.0111,
+      "step": 7210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000243226545092345,
+      "loss": 0.0195,
+      "step": 7211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024321135770084338,
+      "loss": 0.0113,
+      "step": 7212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024319616875253616,
+      "loss": 0.015,
+      "step": 7213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024318097824767697,
+      "loss": 0.0089,
+      "step": 7214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002431657861865196,
+      "loss": 0.0202,
+      "step": 7215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002431505925693177,
+      "loss": 0.0084,
+      "step": 7216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024313539739632506,
+      "loss": 0.0113,
+      "step": 7217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002431202006677955,
+      "loss": 0.0277,
+      "step": 7218
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024310500238398277,
+      "loss": 0.0112,
+      "step": 7219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024308980254514082,
+      "loss": 0.005,
+      "step": 7220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024307460115152341,
+      "loss": 0.0155,
+      "step": 7221
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024305939820338444,
+      "loss": 0.0078,
+      "step": 7222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002430441937009779,
+      "loss": 0.0211,
+      "step": 7223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024302898764455772,
+      "loss": 0.0067,
+      "step": 7224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024301378003437776,
+      "loss": 0.0069,
+      "step": 7225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024299857087069216,
+      "loss": 0.0078,
+      "step": 7226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024298336015375486,
+      "loss": 0.0113,
+      "step": 7227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024296814788381992,
+      "loss": 0.0088,
+      "step": 7228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024295293406114145,
+      "loss": 0.0363,
+      "step": 7229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002429377186859735,
+      "loss": 0.0044,
+      "step": 7230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002429225017585702,
+      "loss": 0.0058,
+      "step": 7231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024290728327918575,
+      "loss": 0.02,
+      "step": 7232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024289206324807426,
+      "loss": 0.0056,
+      "step": 7233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024287684166548999,
+      "loss": 0.0083,
+      "step": 7234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024286161853168716,
+      "loss": 0.0077,
+      "step": 7235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024284639384691996,
+      "loss": 0.0089,
+      "step": 7236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024283116761144276,
+      "loss": 0.0306,
+      "step": 7237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002428159398255098,
+      "loss": 0.0101,
+      "step": 7238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002428007104893755,
+      "loss": 0.0139,
+      "step": 7239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024278547960329412,
+      "loss": 0.0066,
+      "step": 7240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024277024716752007,
+      "loss": 0.0074,
+      "step": 7241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024275501318230783,
+      "loss": 0.0153,
+      "step": 7242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024273977764791176,
+      "loss": 0.0178,
+      "step": 7243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024272454056458635,
+      "loss": 0.0123,
+      "step": 7244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024270930193258603,
+      "loss": 0.0115,
+      "step": 7245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024269406175216542,
+      "loss": 0.015,
+      "step": 7246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024267882002357898,
+      "loss": 0.0156,
+      "step": 7247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024266357674708136,
+      "loss": 0.0211,
+      "step": 7248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024264833192292704,
+      "loss": 0.0033,
+      "step": 7249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024263308555137064,
+      "loss": 0.0056,
+      "step": 7250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002426178376326669,
+      "loss": 0.0092,
+      "step": 7251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002426025881670705,
+      "loss": 0.0052,
+      "step": 7252
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024258733715483603,
+      "loss": 0.0088,
+      "step": 7253
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024257208459621828,
+      "loss": 0.0224,
+      "step": 7254
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024255683049147195,
+      "loss": 0.0147,
+      "step": 7255
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024254157484085185,
+      "loss": 0.0186,
+      "step": 7256
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024252631764461278,
+      "loss": 0.0283,
+      "step": 7257
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002425110589030095,
+      "loss": 0.0078,
+      "step": 7258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024249579861629691,
+      "loss": 0.0058,
+      "step": 7259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024248053678472992,
+      "loss": 0.0099,
+      "step": 7260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002424652734085634,
+      "loss": 0.0231,
+      "step": 7261
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024245000848805226,
+      "loss": 0.0027,
+      "step": 7262
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002424347420234515,
+      "loss": 0.0142,
+      "step": 7263
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024241947401501602,
+      "loss": 0.0301,
+      "step": 7264
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024240420446300088,
+      "loss": 0.0042,
+      "step": 7265
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423889333676611,
+      "loss": 0.0111,
+      "step": 7266
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024237366072925174,
+      "loss": 0.007,
+      "step": 7267
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423583865480279,
+      "loss": 0.0047,
+      "step": 7268
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423431108242446,
+      "loss": 0.0028,
+      "step": 7269
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024232783355815707,
+      "loss": 0.0155,
+      "step": 7270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423125547500205,
+      "loss": 0.0125,
+      "step": 7271
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024229727440008996,
+      "loss": 0.0208,
+      "step": 7272
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422819925086207,
+      "loss": 0.0043,
+      "step": 7273
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000242266709075868,
+      "loss": 0.0086,
+      "step": 7274
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024225142410208708,
+      "loss": 0.0192,
+      "step": 7275
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024223613758753325,
+      "loss": 0.0475,
+      "step": 7276
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422208495324618,
+      "loss": 0.014,
+      "step": 7277
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422055599371281,
+      "loss": 0.0067,
+      "step": 7278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002421902688017875,
+      "loss": 0.0134,
+      "step": 7279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002421749761266954,
+      "loss": 0.0208,
+      "step": 7280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024215968191210717,
+      "loss": 0.0117,
+      "step": 7281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024214438615827828,
+      "loss": 0.0081,
+      "step": 7282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024212908886546425,
+      "loss": 0.0183,
+      "step": 7283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002421137900339205,
+      "loss": 0.005,
+      "step": 7284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002420984896639026,
+      "loss": 0.0225,
+      "step": 7285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024208318775566606,
+      "loss": 0.0066,
+      "step": 7286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024206788430946647,
+      "loss": 0.008,
+      "step": 7287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024205257932555943,
+      "loss": 0.014,
+      "step": 7288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024203727280420058,
+      "loss": 0.0087,
+      "step": 7289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002420219647456455,
+      "loss": 0.0073,
+      "step": 7290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024200665515014988,
+      "loss": 0.0053,
+      "step": 7291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024199134401796946,
+      "loss": 0.0086,
+      "step": 7292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024197603134935999,
+      "loss": 0.0077,
+      "step": 7293
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024196071714457717,
+      "loss": 0.0116,
+      "step": 7294
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024194540140387675,
+      "loss": 0.0312,
+      "step": 7295
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024193008412751456,
+      "loss": 0.0059,
+      "step": 7296
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024191476531574646,
+      "loss": 0.0186,
+      "step": 7297
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024189944496882829,
+      "loss": 0.0109,
+      "step": 7298
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002418841230870159,
+      "loss": 0.0303,
+      "step": 7299
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024186879967056524,
+      "loss": 0.006,
+      "step": 7300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002418534747197322,
+      "loss": 0.0079,
+      "step": 7301
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002418381482347727,
+      "loss": 0.0085,
+      "step": 7302
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024182282021594284,
+      "loss": 0.0018,
+      "step": 7303
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024180749066349852,
+      "loss": 0.0208,
+      "step": 7304
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002417921595776958,
+      "loss": 0.0116,
+      "step": 7305
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024177682695879075,
+      "loss": 0.0054,
+      "step": 7306
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002417614928070394,
+      "loss": 0.0116,
+      "step": 7307
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024174615712269795,
+      "loss": 0.0127,
+      "step": 7308
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024173081990602256,
+      "loss": 0.0106,
+      "step": 7309
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024171548115726923,
+      "loss": 0.0227,
+      "step": 7310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024170014087669426,
+      "loss": 0.0505,
+      "step": 7311
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024168479906455385,
+      "loss": 0.0163,
+      "step": 7312
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024166945572110427,
+      "loss": 0.0064,
+      "step": 7313
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024165411084660167,
+      "loss": 0.0121,
+      "step": 7314
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024163876444130246,
+      "loss": 0.0981,
+      "step": 7315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024162341650546285,
+      "loss": 0.0258,
+      "step": 7316
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024160806703933928,
+      "loss": 0.0061,
+      "step": 7317
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024159271604318804,
+      "loss": 0.0093,
+      "step": 7318
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024157736351726556,
+      "loss": 0.0118,
+      "step": 7319
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024156200946182828,
+      "loss": 0.02,
+      "step": 7320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024154665387713257,
+      "loss": 0.011,
+      "step": 7321
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024153129676343491,
+      "loss": 0.0197,
+      "step": 7322
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002415159381209919,
+      "loss": 0.0213,
+      "step": 7323
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002415005779500599,
+      "loss": 0.0135,
+      "step": 7324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024148521625089552,
+      "loss": 0.0279,
+      "step": 7325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024146985302375538,
+      "loss": 0.0097,
+      "step": 7326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024145448826889604,
+      "loss": 0.005,
+      "step": 7327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024143912198657406,
+      "loss": 0.0014,
+      "step": 7328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002414237541770462,
+      "loss": 0.0094,
+      "step": 7329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024140838484056903,
+      "loss": 0.0048,
+      "step": 7330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002413930139773993,
+      "loss": 0.0073,
+      "step": 7331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024137764158779372,
+      "loss": 0.0146,
+      "step": 7332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024136226767200903,
+      "loss": 0.0205,
+      "step": 7333
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024134689223030201,
+      "loss": 0.0033,
+      "step": 7334
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024133151526292948,
+      "loss": 0.0279,
+      "step": 7335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024131613677014826,
+      "loss": 0.0219,
+      "step": 7336
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024130075675221513,
+      "loss": 0.0121,
+      "step": 7337
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024128537520938704,
+      "loss": 0.0499,
+      "step": 7338
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024126999214192088,
+      "loss": 0.0107,
+      "step": 7339
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002412546075500736,
+      "loss": 0.0084,
+      "step": 7340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002412392214341021,
+      "loss": 0.0036,
+      "step": 7341
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024122383379426333,
+      "loss": 0.0282,
+      "step": 7342
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024120844463081437,
+      "loss": 0.0145,
+      "step": 7343
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024119305394401221,
+      "loss": 0.0337,
+      "step": 7344
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024117766173411395,
+      "loss": 0.0083,
+      "step": 7345
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024116226800137658,
+      "loss": 0.031,
+      "step": 7346
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024114687274605725,
+      "loss": 0.0125,
+      "step": 7347
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024113147596841313,
+      "loss": 0.0154,
+      "step": 7348
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024111607766870132,
+      "loss": 0.0071,
+      "step": 7349
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024110067784717903,
+      "loss": 0.0163,
+      "step": 7350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024108527650410343,
+      "loss": 0.0229,
+      "step": 7351
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002410698736397318,
+      "loss": 0.0252,
+      "step": 7352
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024105446925432137,
+      "loss": 0.0038,
+      "step": 7353
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024103906334812942,
+      "loss": 0.0217,
+      "step": 7354
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024102365592141325,
+      "loss": 0.0198,
+      "step": 7355
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024100824697443024,
+      "loss": 0.0083,
+      "step": 7356
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024099283650743769,
+      "loss": 0.0094,
+      "step": 7357
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000240977424520693,
+      "loss": 0.0116,
+      "step": 7358
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024096201101445362,
+      "loss": 0.0127,
+      "step": 7359
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024094659598897694,
+      "loss": 0.0155,
+      "step": 7360
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024093117944452041,
+      "loss": 0.0078,
+      "step": 7361
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024091576138134157,
+      "loss": 0.0167,
+      "step": 7362
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002409003417996979,
+      "loss": 0.0313,
+      "step": 7363
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002408849206998469,
+      "loss": 0.0487,
+      "step": 7364
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002408694980820462,
+      "loss": 0.0089,
+      "step": 7365
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024085407394655334,
+      "loss": 0.0116,
+      "step": 7366
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024083864829362592,
+      "loss": 0.0056,
+      "step": 7367
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024082322112352168,
+      "loss": 0.0075,
+      "step": 7368
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024080779243649815,
+      "loss": 0.0055,
+      "step": 7369
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024079236223281306,
+      "loss": 0.0176,
+      "step": 7370
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024077693051272416,
+      "loss": 0.007,
+      "step": 7371
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024076149727648917,
+      "loss": 0.0273,
+      "step": 7372
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002407460625243659,
+      "loss": 0.0226,
+      "step": 7373
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000240730626256612,
+      "loss": 0.0042,
+      "step": 7374
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024071518847348542,
+      "loss": 0.0126,
+      "step": 7375
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024069974917524396,
+      "loss": 0.0086,
+      "step": 7376
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024068430836214548,
+      "loss": 0.0074,
+      "step": 7377
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024066886603444786,
+      "loss": 0.0062,
+      "step": 7378
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024065342219240903,
+      "loss": 0.0052,
+      "step": 7379
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002406379768362869,
+      "loss": 0.0106,
+      "step": 7380
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002406225299663395,
+      "loss": 0.008,
+      "step": 7381
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002406070815828248,
+      "loss": 0.0172,
+      "step": 7382
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024059163168600078,
+      "loss": 0.0098,
+      "step": 7383
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024057618027612554,
+      "loss": 0.0371,
+      "step": 7384
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024056072735345707,
+      "loss": 0.0251,
+      "step": 7385
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002405452729182535,
+      "loss": 0.0086,
+      "step": 7386
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024052981697077303,
+      "loss": 0.0189,
+      "step": 7387
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002405143595112737,
+      "loss": 0.0328,
+      "step": 7388
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002404989005400137,
+      "loss": 0.0029,
+      "step": 7389
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024048344005725119,
+      "loss": 0.013,
+      "step": 7390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024046797806324445,
+      "loss": 0.0065,
+      "step": 7391
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024045251455825174,
+      "loss": 0.0167,
+      "step": 7392
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024043704954253127,
+      "loss": 0.0074,
+      "step": 7393
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024042158301634136,
+      "loss": 0.0088,
+      "step": 7394
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024040611497994037,
+      "loss": 0.0225,
+      "step": 7395
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024039064543358653,
+      "loss": 0.0043,
+      "step": 7396
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024037517437753832,
+      "loss": 0.0145,
+      "step": 7397
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024035970181205416,
+      "loss": 0.0189,
+      "step": 7398
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024034422773739231,
+      "loss": 0.0286,
+      "step": 7399
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024032875215381142,
+      "loss": 0.0139,
+      "step": 7400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024031327506156982,
+      "loss": 0.0245,
+      "step": 7401
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000240297796460926,
+      "loss": 0.027,
+      "step": 7402
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002402823163521386,
+      "loss": 0.0196,
+      "step": 7403
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024026683473546607,
+      "loss": 0.0221,
+      "step": 7404
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024025135161116702,
+      "loss": 0.0054,
+      "step": 7405
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024023586697950007,
+      "loss": 0.0132,
+      "step": 7406
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024022038084072373,
+      "loss": 0.0179,
+      "step": 7407
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024020489319509674,
+      "loss": 0.0091,
+      "step": 7408
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002401894040428778,
+      "loss": 0.0087,
+      "step": 7409
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024017391338432558,
+      "loss": 0.0107,
+      "step": 7410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024015842121969874,
+      "loss": 0.0081,
+      "step": 7411
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024014292754925612,
+      "loss": 0.0077,
+      "step": 7412
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024012743237325642,
+      "loss": 0.0073,
+      "step": 7413
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002401119356919585,
+      "loss": 0.029,
+      "step": 7414
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024009643750562114,
+      "loss": 0.0035,
+      "step": 7415
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024008093781450324,
+      "loss": 0.0228,
+      "step": 7416
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002400654366188636,
+      "loss": 0.0115,
+      "step": 7417
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024004993391896115,
+      "loss": 0.0198,
+      "step": 7418
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024003442971505485,
+      "loss": 0.0266,
+      "step": 7419
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002400189240074036,
+      "loss": 0.0123,
+      "step": 7420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002400034167962664,
+      "loss": 0.0183,
+      "step": 7421
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023998790808190225,
+      "loss": 0.0184,
+      "step": 7422
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023997239786457017,
+      "loss": 0.0045,
+      "step": 7423
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023995688614452922,
+      "loss": 0.0139,
+      "step": 7424
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023994137292203846,
+      "loss": 0.0252,
+      "step": 7425
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000239925858197357,
+      "loss": 0.024,
+      "step": 7426
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023991034197074393,
+      "loss": 0.0239,
+      "step": 7427
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023989482424245848,
+      "loss": 0.0221,
+      "step": 7428
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023987930501275978,
+      "loss": 0.0208,
+      "step": 7429
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000239863784281907,
+      "loss": 0.0111,
+      "step": 7430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023984826205015937,
+      "loss": 0.0101,
+      "step": 7431
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023983273831777618,
+      "loss": 0.0188,
+      "step": 7432
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002398172130850167,
+      "loss": 0.0224,
+      "step": 7433
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002398016863521402,
+      "loss": 0.0156,
+      "step": 7434
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023978615811940605,
+      "loss": 0.0292,
+      "step": 7435
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002397706283870736,
+      "loss": 0.027,
+      "step": 7436
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023975509715540216,
+      "loss": 0.0102,
+      "step": 7437
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023973956442465116,
+      "loss": 0.0087,
+      "step": 7438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023972403019508008,
+      "loss": 0.0229,
+      "step": 7439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023970849446694825,
+      "loss": 0.0059,
+      "step": 7440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023969295724051532,
+      "loss": 0.0166,
+      "step": 7441
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023967741851604066,
+      "loss": 0.0052,
+      "step": 7442
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023966187829378385,
+      "loss": 0.0497,
+      "step": 7443
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023964633657400441,
+      "loss": 0.0232,
+      "step": 7444
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002396307933569619,
+      "loss": 0.0197,
+      "step": 7445
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023961524864291598,
+      "loss": 0.0121,
+      "step": 7446
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023959970243212625,
+      "loss": 0.0053,
+      "step": 7447
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023958415472485236,
+      "loss": 0.0054,
+      "step": 7448
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023956860552135396,
+      "loss": 0.0052,
+      "step": 7449
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023955305482189083,
+      "loss": 0.0015,
+      "step": 7450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002395375026267226,
+      "loss": 0.0214,
+      "step": 7451
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023952194893610905,
+      "loss": 0.0325,
+      "step": 7452
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023950639375030998,
+      "loss": 0.022,
+      "step": 7453
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023949083706958517,
+      "loss": 0.0046,
+      "step": 7454
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002394752788941945,
+      "loss": 0.0518,
+      "step": 7455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023945971922439776,
+      "loss": 0.0194,
+      "step": 7456
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023944415806045481,
+      "loss": 0.0251,
+      "step": 7457
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023942859540262558,
+      "loss": 0.0057,
+      "step": 7458
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023941303125117,
+      "loss": 0.0273,
+      "step": 7459
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023939746560634804,
+      "loss": 0.0209,
+      "step": 7460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023938189846841966,
+      "loss": 0.0198,
+      "step": 7461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023936632983764485,
+      "loss": 0.0194,
+      "step": 7462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002393507597142836,
+      "loss": 0.004,
+      "step": 7463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023933518809859608,
+      "loss": 0.0139,
+      "step": 7464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023931961499084226,
+      "loss": 0.0128,
+      "step": 7465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023930404039128222,
+      "loss": 0.0197,
+      "step": 7466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023928846430017618,
+      "loss": 0.0235,
+      "step": 7467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023927288671778422,
+      "loss": 0.0069,
+      "step": 7468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023925730764436651,
+      "loss": 0.0035,
+      "step": 7469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023924172708018334,
+      "loss": 0.014,
+      "step": 7470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023922614502549486,
+      "loss": 0.0149,
+      "step": 7471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002392105614805613,
+      "loss": 0.0136,
+      "step": 7472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023919497644564298,
+      "loss": 0.0201,
+      "step": 7473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002391793899210002,
+      "loss": 0.0122,
+      "step": 7474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023916380190689327,
+      "loss": 0.0118,
+      "step": 7475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002391482124035825,
+      "loss": 0.0246,
+      "step": 7476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023913262141132832,
+      "loss": 0.0095,
+      "step": 7477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023911702893039113,
+      "loss": 0.0163,
+      "step": 7478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023910143496103136,
+      "loss": 0.0175,
+      "step": 7479
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023908583950350937,
+      "loss": 0.005,
+      "step": 7480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002390702425580857,
+      "loss": 0.0254,
+      "step": 7481
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023905464412502092,
+      "loss": 0.023,
+      "step": 7482
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023903904420457544,
+      "loss": 0.0058,
+      "step": 7483
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023902344279700985,
+      "loss": 0.0477,
+      "step": 7484
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002390078399025847,
+      "loss": 0.015,
+      "step": 7485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023899223552156063,
+      "loss": 0.0036,
+      "step": 7486
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023897662965419826,
+      "loss": 0.0078,
+      "step": 7487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002389610223007582,
+      "loss": 0.018,
+      "step": 7488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023894541346150116,
+      "loss": 0.0398,
+      "step": 7489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002389298031366878,
+      "loss": 0.009,
+      "step": 7490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023891419132657886,
+      "loss": 0.007,
+      "step": 7491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023889857803143508,
+      "loss": 0.0052,
+      "step": 7492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023888296325151728,
+      "loss": 0.0183,
+      "step": 7493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023886734698708617,
+      "loss": 0.0036,
+      "step": 7494
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002388517292384026,
+      "loss": 0.0061,
+      "step": 7495
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023883611000572748,
+      "loss": 0.017,
+      "step": 7496
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023882048928932162,
+      "loss": 0.0215,
+      "step": 7497
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023880486708944594,
+      "loss": 0.0085,
+      "step": 7498
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023878924340636133,
+      "loss": 0.0071,
+      "step": 7499
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023877361824032874,
+      "loss": 0.0214,
+      "step": 7500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023875799159160917,
+      "loss": 0.0147,
+      "step": 7501
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023874236346046362,
+      "loss": 0.0094,
+      "step": 7502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023872673384715305,
+      "loss": 0.0106,
+      "step": 7503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023871110275193856,
+      "loss": 0.0031,
+      "step": 7504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023869547017508116,
+      "loss": 0.0237,
+      "step": 7505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023867983611684202,
+      "loss": 0.0299,
+      "step": 7506
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023866420057748225,
+      "loss": 0.0232,
+      "step": 7507
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023864856355726294,
+      "loss": 0.0157,
+      "step": 7508
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023863292505644528,
+      "loss": 0.0322,
+      "step": 7509
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023861728507529044,
+      "loss": 0.0047,
+      "step": 7510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023860164361405974,
+      "loss": 0.0073,
+      "step": 7511
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023858600067301427,
+      "loss": 0.0057,
+      "step": 7512
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023857035625241539,
+      "loss": 0.0222,
+      "step": 7513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023855471035252437,
+      "loss": 0.0054,
+      "step": 7514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023853906297360256,
+      "loss": 0.0052,
+      "step": 7515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002385234141159112,
+      "loss": 0.0195,
+      "step": 7516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023850776377971178,
+      "loss": 0.0117,
+      "step": 7517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023849211196526562,
+      "loss": 0.0071,
+      "step": 7518
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023847645867283411,
+      "loss": 0.0173,
+      "step": 7519
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002384608039026788,
+      "loss": 0.0063,
+      "step": 7520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023844514765506105,
+      "loss": 0.0085,
+      "step": 7521
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002384294899302424,
+      "loss": 0.0045,
+      "step": 7522
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023841383072848433,
+      "loss": 0.0296,
+      "step": 7523
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023839817005004836,
+      "loss": 0.0564,
+      "step": 7524
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023838250789519615,
+      "loss": 0.0073,
+      "step": 7525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023836684426418916,
+      "loss": 0.0383,
+      "step": 7526
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002383511791572891,
+      "loss": 0.0427,
+      "step": 7527
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023833551257475757,
+      "loss": 0.017,
+      "step": 7528
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023831984451685626,
+      "loss": 0.0088,
+      "step": 7529
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023830417498384683,
+      "loss": 0.006,
+      "step": 7530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023828850397599098,
+      "loss": 0.0028,
+      "step": 7531
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002382728314935504,
+      "loss": 0.0322,
+      "step": 7532
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000238257157536787,
+      "loss": 0.0119,
+      "step": 7533
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023824148210596242,
+      "loss": 0.0084,
+      "step": 7534
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023822580520133852,
+      "loss": 0.0036,
+      "step": 7535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023821012682317716,
+      "loss": 0.0135,
+      "step": 7536
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023819444697174014,
+      "loss": 0.051,
+      "step": 7537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002381787656472894,
+      "loss": 0.0152,
+      "step": 7538
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023816308285008685,
+      "loss": 0.014,
+      "step": 7539
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023814739858039432,
+      "loss": 0.0175,
+      "step": 7540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023813171283847392,
+      "loss": 0.0204,
+      "step": 7541
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002381160256245875,
+      "loss": 0.0167,
+      "step": 7542
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023810033693899717,
+      "loss": 0.011,
+      "step": 7543
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023808464678196488,
+      "loss": 0.0267,
+      "step": 7544
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023806895515375275,
+      "loss": 0.0114,
+      "step": 7545
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023805326205462277,
+      "loss": 0.0273,
+      "step": 7546
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023803756748483711,
+      "loss": 0.0115,
+      "step": 7547
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023802187144465796,
+      "loss": 0.0318,
+      "step": 7548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023800617393434736,
+      "loss": 0.0134,
+      "step": 7549
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023799047495416754,
+      "loss": 0.0231,
+      "step": 7550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023797477450438067,
+      "loss": 0.0082,
+      "step": 7551
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000237959072585249,
+      "loss": 0.0149,
+      "step": 7552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023794336919703482,
+      "loss": 0.0143,
+      "step": 7553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023792766434000036,
+      "loss": 0.0073,
+      "step": 7554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002379119580144079,
+      "loss": 0.0155,
+      "step": 7555
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023789625022051984,
+      "loss": 0.0038,
+      "step": 7556
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002378805409585984,
+      "loss": 0.017,
+      "step": 7557
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023786483022890614,
+      "loss": 0.0187,
+      "step": 7558
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002378491180317053,
+      "loss": 0.0223,
+      "step": 7559
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023783340436725838,
+      "loss": 0.0096,
+      "step": 7560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002378176892358278,
+      "loss": 0.0038,
+      "step": 7561
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002378019726376761,
+      "loss": 0.0306,
+      "step": 7562
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023778625457306565,
+      "loss": 0.0236,
+      "step": 7563
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023777053504225908,
+      "loss": 0.0219,
+      "step": 7564
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023775481404551886,
+      "loss": 0.0169,
+      "step": 7565
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002377390915831076,
+      "loss": 0.0175,
+      "step": 7566
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023772336765528796,
+      "loss": 0.0105,
+      "step": 7567
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002377076422623224,
+      "loss": 0.0121,
+      "step": 7568
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023769191540447368,
+      "loss": 0.0139,
+      "step": 7569
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023767618708200444,
+      "loss": 0.0079,
+      "step": 7570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023766045729517742,
+      "loss": 0.0128,
+      "step": 7571
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023764472604425525,
+      "loss": 0.0054,
+      "step": 7572
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023762899332950072,
+      "loss": 0.012,
+      "step": 7573
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023761325915117658,
+      "loss": 0.0109,
+      "step": 7574
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023759752350954565,
+      "loss": 0.0192,
+      "step": 7575
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023758178640487075,
+      "loss": 0.01,
+      "step": 7576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023756604783741468,
+      "loss": 0.0039,
+      "step": 7577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023755030780744026,
+      "loss": 0.0188,
+      "step": 7578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023753456631521049,
+      "loss": 0.0075,
+      "step": 7579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002375188233609882,
+      "loss": 0.0185,
+      "step": 7580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002375030789450364,
+      "loss": 0.0211,
+      "step": 7581
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000237487333067618,
+      "loss": 0.0174,
+      "step": 7582
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023747158572899596,
+      "loss": 0.0111,
+      "step": 7583
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023745583692943333,
+      "loss": 0.0268,
+      "step": 7584
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023744008666919317,
+      "loss": 0.0202,
+      "step": 7585
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023742433494853846,
+      "loss": 0.0198,
+      "step": 7586
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002374085817677324,
+      "loss": 0.0118,
+      "step": 7587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023739282712703796,
+      "loss": 0.0041,
+      "step": 7588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023737707102671838,
+      "loss": 0.0251,
+      "step": 7589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023736131346703679,
+      "loss": 0.0051,
+      "step": 7590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023734555444825632,
+      "loss": 0.0183,
+      "step": 7591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023732979397064026,
+      "loss": 0.0189,
+      "step": 7592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023731403203445175,
+      "loss": 0.0102,
+      "step": 7593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002372982686399541,
+      "loss": 0.0054,
+      "step": 7594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023728250378741056,
+      "loss": 0.0045,
+      "step": 7595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023726673747708447,
+      "loss": 0.0196,
+      "step": 7596
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023725096970923914,
+      "loss": 0.0049,
+      "step": 7597
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002372352004841379,
+      "loss": 0.0667,
+      "step": 7598
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023721942980204418,
+      "loss": 0.0034,
+      "step": 7599
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023720365766322134,
+      "loss": 0.0274,
+      "step": 7600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023718788406793274,
+      "loss": 0.0335,
+      "step": 7601
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000237172109016442,
+      "loss": 0.0041,
+      "step": 7602
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002371563325090124,
+      "loss": 0.0161,
+      "step": 7603
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023714055454590756,
+      "loss": 0.0028,
+      "step": 7604
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000237124775127391,
+      "loss": 0.0083,
+      "step": 7605
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023710899425372618,
+      "loss": 0.0127,
+      "step": 7606
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023709321192517678,
+      "loss": 0.0188,
+      "step": 7607
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002370774281420063,
+      "loss": 0.0092,
+      "step": 7608
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023706164290447843,
+      "loss": 0.0075,
+      "step": 7609
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023704585621285676,
+      "loss": 0.0185,
+      "step": 7610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000237030068067405,
+      "loss": 0.0018,
+      "step": 7611
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023701427846838682,
+      "loss": 0.0078,
+      "step": 7612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023699848741606596,
+      "loss": 0.0255,
+      "step": 7613
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023698269491070607,
+      "loss": 0.0493,
+      "step": 7614
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023696690095257106,
+      "loss": 0.0298,
+      "step": 7615
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023695110554192457,
+      "loss": 0.0219,
+      "step": 7616
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002369353086790305,
+      "loss": 0.0052,
+      "step": 7617
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023691951036415273,
+      "loss": 0.0207,
+      "step": 7618
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000236903710597555,
+      "loss": 0.0061,
+      "step": 7619
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002368879093795013,
+      "loss": 0.0085,
+      "step": 7620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023687210671025552,
+      "loss": 0.0069,
+      "step": 7621
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023685630259008157,
+      "loss": 0.0179,
+      "step": 7622
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023684049701924338,
+      "loss": 0.0039,
+      "step": 7623
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023682468999800498,
+      "loss": 0.003,
+      "step": 7624
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002368088815266303,
+      "loss": 0.0066,
+      "step": 7625
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023679307160538352,
+      "loss": 0.0065,
+      "step": 7626
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002367772602345286,
+      "loss": 0.0426,
+      "step": 7627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023676144741432962,
+      "loss": 0.0256,
+      "step": 7628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023674563314505067,
+      "loss": 0.0238,
+      "step": 7629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023672981742695593,
+      "loss": 0.0141,
+      "step": 7630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023671400026030953,
+      "loss": 0.024,
+      "step": 7631
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002366981816453756,
+      "loss": 0.0275,
+      "step": 7632
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023668236158241843,
+      "loss": 0.0078,
+      "step": 7633
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002366665400717022,
+      "loss": 0.0094,
+      "step": 7634
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023665071711349113,
+      "loss": 0.031,
+      "step": 7635
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023663489270804954,
+      "loss": 0.038,
+      "step": 7636
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002366190668556417,
+      "loss": 0.0086,
+      "step": 7637
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002366032395565319,
+      "loss": 0.0085,
+      "step": 7638
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023658741081098459,
+      "loss": 0.0291,
+      "step": 7639
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023657158061926397,
+      "loss": 0.0271,
+      "step": 7640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023655574898163466,
+      "loss": 0.0126,
+      "step": 7641
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002365399158983609,
+      "loss": 0.0167,
+      "step": 7642
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023652408136970717,
+      "loss": 0.0371,
+      "step": 7643
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000236508245395938,
+      "loss": 0.0144,
+      "step": 7644
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002364924079773178,
+      "loss": 0.0412,
+      "step": 7645
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023647656911411112,
+      "loss": 0.0192,
+      "step": 7646
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023646072880658254,
+      "loss": 0.0194,
+      "step": 7647
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023644488705499653,
+      "loss": 0.0084,
+      "step": 7648
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023642904385961778,
+      "loss": 0.0045,
+      "step": 7649
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002364131992207108,
+      "loss": 0.0045,
+      "step": 7650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002363973531385403,
+      "loss": 0.0029,
+      "step": 7651
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023638150561337094,
+      "loss": 0.0186,
+      "step": 7652
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023636565664546735,
+      "loss": 0.0176,
+      "step": 7653
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023634980623509427,
+      "loss": 0.0068,
+      "step": 7654
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023633395438251643,
+      "loss": 0.027,
+      "step": 7655
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023631810108799863,
+      "loss": 0.0077,
+      "step": 7656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023630224635180557,
+      "loss": 0.0089,
+      "step": 7657
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023628639017420205,
+      "loss": 0.0194,
+      "step": 7658
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000236270532555453,
+      "loss": 0.0137,
+      "step": 7659
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002362546734958232,
+      "loss": 0.0147,
+      "step": 7660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023623881299557749,
+      "loss": 0.0159,
+      "step": 7661
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023622295105498082,
+      "loss": 0.0073,
+      "step": 7662
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023620708767429816,
+      "loss": 0.012,
+      "step": 7663
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023619122285379438,
+      "loss": 0.0129,
+      "step": 7664
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023617535659373445,
+      "loss": 0.0097,
+      "step": 7665
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023615948889438345,
+      "loss": 0.0348,
+      "step": 7666
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023614361975600633,
+      "loss": 0.0072,
+      "step": 7667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023612774917886815,
+      "loss": 0.0341,
+      "step": 7668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000236111877163234,
+      "loss": 0.0381,
+      "step": 7669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023609600370936895,
+      "loss": 0.0134,
+      "step": 7670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023608012881753807,
+      "loss": 0.0136,
+      "step": 7671
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023606425248800654,
+      "loss": 0.0058,
+      "step": 7672
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360483747210396,
+      "loss": 0.0147,
+      "step": 7673
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360324955169024,
+      "loss": 0.0199,
+      "step": 7674
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023601661487586008,
+      "loss": 0.0181,
+      "step": 7675
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360007327981779,
+      "loss": 0.0185,
+      "step": 7676
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023598484928412115,
+      "loss": 0.011,
+      "step": 7677
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002359689643339552,
+      "loss": 0.0102,
+      "step": 7678
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023595307794794515,
+      "loss": 0.015,
+      "step": 7679
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023593719012635647,
+      "loss": 0.0265,
+      "step": 7680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023592130086945453,
+      "loss": 0.0101,
+      "step": 7681
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023590541017750469,
+      "loss": 0.007,
+      "step": 7682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023588951805077231,
+      "loss": 0.0154,
+      "step": 7683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002358736244895229,
+      "loss": 0.0148,
+      "step": 7684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023585772949402182,
+      "loss": 0.0184,
+      "step": 7685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023584183306453457,
+      "loss": 0.0085,
+      "step": 7686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023582593520132673,
+      "loss": 0.0092,
+      "step": 7687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023581003590466372,
+      "loss": 0.0109,
+      "step": 7688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023579413517481115,
+      "loss": 0.0184,
+      "step": 7689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023577823301203457,
+      "loss": 0.02,
+      "step": 7690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023576232941659956,
+      "loss": 0.0136,
+      "step": 7691
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002357464243887718,
+      "loss": 0.0062,
+      "step": 7692
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023573051792881685,
+      "loss": 0.0029,
+      "step": 7693
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023571461003700047,
+      "loss": 0.0014,
+      "step": 7694
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023569870071358826,
+      "loss": 0.0181,
+      "step": 7695
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023568278995884603,
+      "loss": 0.0038,
+      "step": 7696
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023566687777303944,
+      "loss": 0.0101,
+      "step": 7697
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002356509641564343,
+      "loss": 0.0061,
+      "step": 7698
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023563504910929634,
+      "loss": 0.0178,
+      "step": 7699
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023561913263189143,
+      "loss": 0.0122,
+      "step": 7700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023560321472448543,
+      "loss": 0.0235,
+      "step": 7701
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002355872953873441,
+      "loss": 0.0286,
+      "step": 7702
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002355713746207334,
+      "loss": 0.0147,
+      "step": 7703
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023555545242491926,
+      "loss": 0.0117,
+      "step": 7704
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023553952880016747,
+      "loss": 0.0167,
+      "step": 7705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023552360374674418,
+      "loss": 0.0083,
+      "step": 7706
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023550767726491522,
+      "loss": 0.0178,
+      "step": 7707
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023549174935494666,
+      "loss": 0.0069,
+      "step": 7708
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023547582001710448,
+      "loss": 0.0239,
+      "step": 7709
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002354598892516548,
+      "loss": 0.0268,
+      "step": 7710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002354439570588636,
+      "loss": 0.0183,
+      "step": 7711
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023542802343899706,
+      "loss": 0.0036,
+      "step": 7712
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023541208839232127,
+      "loss": 0.0073,
+      "step": 7713
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023539615191910237,
+      "loss": 0.011,
+      "step": 7714
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353802140196066,
+      "loss": 0.0089,
+      "step": 7715
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353642746941,
+      "loss": 0.0155,
+      "step": 7716
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023534833394284893,
+      "loss": 0.0187,
+      "step": 7717
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023533239176611953,
+      "loss": 0.0193,
+      "step": 7718
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353164481641781,
+      "loss": 0.0066,
+      "step": 7719
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023530050313729105,
+      "loss": 0.0103,
+      "step": 7720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023528455668572453,
+      "loss": 0.0145,
+      "step": 7721
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023526860880974484,
+      "loss": 0.011,
+      "step": 7722
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023525265950961852,
+      "loss": 0.0058,
+      "step": 7723
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023523670878561185,
+      "loss": 0.0148,
+      "step": 7724
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023522075663799124,
+      "loss": 0.0129,
+      "step": 7725
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002352048030670231,
+      "loss": 0.0231,
+      "step": 7726
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023518884807297396,
+      "loss": 0.0055,
+      "step": 7727
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002351728916561102,
+      "loss": 0.0169,
+      "step": 7728
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002351569338166984,
+      "loss": 0.0111,
+      "step": 7729
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023514097455500508,
+      "loss": 0.0166,
+      "step": 7730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023512501387129675,
+      "loss": 0.0026,
+      "step": 7731
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023510905176584002,
+      "loss": 0.0069,
+      "step": 7732
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023509308823890145,
+      "loss": 0.0073,
+      "step": 7733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023507712329074766,
+      "loss": 0.0135,
+      "step": 7734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023506115692164536,
+      "loss": 0.0209,
+      "step": 7735
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023504518913186116,
+      "loss": 0.0036,
+      "step": 7736
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023502921992166176,
+      "loss": 0.0243,
+      "step": 7737
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002350132492913139,
+      "loss": 0.0168,
+      "step": 7738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023499727724108428,
+      "loss": 0.0218,
+      "step": 7739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002349813037712397,
+      "loss": 0.0543,
+      "step": 7740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002349653288820469,
+      "loss": 0.0105,
+      "step": 7741
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023494935257377278,
+      "loss": 0.0264,
+      "step": 7742
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002349333748466841,
+      "loss": 0.0183,
+      "step": 7743
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023491739570104773,
+      "loss": 0.0153,
+      "step": 7744
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023490141513713057,
+      "loss": 0.005,
+      "step": 7745
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002348854331551995,
+      "loss": 0.0091,
+      "step": 7746
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023486944975552147,
+      "loss": 0.0088,
+      "step": 7747
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023485346493836343,
+      "loss": 0.0127,
+      "step": 7748
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023483747870399233,
+      "loss": 0.0115,
+      "step": 7749
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023482149105267521,
+      "loss": 0.0264,
+      "step": 7750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002348055019846791,
+      "loss": 0.0269,
+      "step": 7751
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023478951150027104,
+      "loss": 0.0427,
+      "step": 7752
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023477351959971804,
+      "loss": 0.0111,
+      "step": 7753
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023475752628328723,
+      "loss": 0.0125,
+      "step": 7754
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023474153155124574,
+      "loss": 0.019,
+      "step": 7755
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023472553540386073,
+      "loss": 0.0122,
+      "step": 7756
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002347095378413994,
+      "loss": 0.0028,
+      "step": 7757
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002346935388641288,
+      "loss": 0.0023,
+      "step": 7758
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002346775384723163,
+      "loss": 0.014,
+      "step": 7759
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000234661536666229,
+      "loss": 0.0042,
+      "step": 7760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023464553344613428,
+      "loss": 0.0185,
+      "step": 7761
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002346295288122994,
+      "loss": 0.0357,
+      "step": 7762
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023461352276499156,
+      "loss": 0.0086,
+      "step": 7763
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002345975153044782,
+      "loss": 0.0082,
+      "step": 7764
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023458150643102665,
+      "loss": 0.035,
+      "step": 7765
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023456549614490432,
+      "loss": 0.0098,
+      "step": 7766
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023454948444637854,
+      "loss": 0.0182,
+      "step": 7767
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023453347133571673,
+      "loss": 0.014,
+      "step": 7768
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023451745681318646,
+      "loss": 0.0045,
+      "step": 7769
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023450144087905508,
+      "loss": 0.007,
+      "step": 7770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023448542353359013,
+      "loss": 0.0047,
+      "step": 7771
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023446940477705913,
+      "loss": 0.0244,
+      "step": 7772
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023445338460972962,
+      "loss": 0.018,
+      "step": 7773
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023443736303186918,
+      "loss": 0.0157,
+      "step": 7774
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002344213400437454,
+      "loss": 0.0157,
+      "step": 7775
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023440531564562588,
+      "loss": 0.0153,
+      "step": 7776
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023438928983777826,
+      "loss": 0.0106,
+      "step": 7777
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002343732626204702,
+      "loss": 0.0111,
+      "step": 7778
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002343572339939694,
+      "loss": 0.054,
+      "step": 7779
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023434120395854355,
+      "loss": 0.0081,
+      "step": 7780
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023432517251446036,
+      "loss": 0.032,
+      "step": 7781
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023430913966198766,
+      "loss": 0.0216,
+      "step": 7782
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023429310540139313,
+      "loss": 0.0135,
+      "step": 7783
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002342770697329447,
+      "loss": 0.0226,
+      "step": 7784
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023426103265691004,
+      "loss": 0.0079,
+      "step": 7785
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023424499417355713,
+      "loss": 0.0276,
+      "step": 7786
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023422895428315378,
+      "loss": 0.0029,
+      "step": 7787
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023421291298596786,
+      "loss": 0.0099,
+      "step": 7788
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002341968702822674,
+      "loss": 0.0185,
+      "step": 7789
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023418082617232026,
+      "loss": 0.0197,
+      "step": 7790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002341647806563944,
+      "loss": 0.0158,
+      "step": 7791
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023414873373475784,
+      "loss": 0.0135,
+      "step": 7792
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023413268540767857,
+      "loss": 0.0086,
+      "step": 7793
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023411663567542466,
+      "loss": 0.0237,
+      "step": 7794
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023410058453826414,
+      "loss": 0.0112,
+      "step": 7795
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023408453199646507,
+      "loss": 0.0076,
+      "step": 7796
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023406847805029565,
+      "loss": 0.017,
+      "step": 7797
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000234052422700024,
+      "loss": 0.0178,
+      "step": 7798
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023403636594591815,
+      "loss": 0.0131,
+      "step": 7799
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002340203077882464,
+      "loss": 0.0056,
+      "step": 7800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023400424822727687,
+      "loss": 0.0083,
+      "step": 7801
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023398818726327785,
+      "loss": 0.0095,
+      "step": 7802
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002339721248965176,
+      "loss": 0.0064,
+      "step": 7803
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023395606112726428,
+      "loss": 0.0077,
+      "step": 7804
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002339399959557863,
+      "loss": 0.0065,
+      "step": 7805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023392392938235195,
+      "loss": 0.0065,
+      "step": 7806
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023390786140722953,
+      "loss": 0.0266,
+      "step": 7807
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023389179203068747,
+      "loss": 0.0176,
+      "step": 7808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023387572125299413,
+      "loss": 0.01,
+      "step": 7809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023385964907441788,
+      "loss": 0.0124,
+      "step": 7810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023384357549522723,
+      "loss": 0.0121,
+      "step": 7811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023382750051569062,
+      "loss": 0.0464,
+      "step": 7812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023381142413607647,
+      "loss": 0.007,
+      "step": 7813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023379534635665334,
+      "loss": 0.0217,
+      "step": 7814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023377926717768973,
+      "loss": 0.005,
+      "step": 7815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023376318659945426,
+      "loss": 0.0385,
+      "step": 7816
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023374710462221543,
+      "loss": 0.0035,
+      "step": 7817
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002337310212462419,
+      "loss": 0.0051,
+      "step": 7818
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023371493647180222,
+      "loss": 0.0137,
+      "step": 7819
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023369885029916507,
+      "loss": 0.0108,
+      "step": 7820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336827627285992,
+      "loss": 0.0134,
+      "step": 7821
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336666737603732,
+      "loss": 0.0525,
+      "step": 7822
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336505833947558,
+      "loss": 0.0034,
+      "step": 7823
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023363449163201576,
+      "loss": 0.0202,
+      "step": 7824
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336183984724218,
+      "loss": 0.0158,
+      "step": 7825
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023360230391624283,
+      "loss": 0.0042,
+      "step": 7826
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002335862079637475,
+      "loss": 0.0105,
+      "step": 7827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023357011061520474,
+      "loss": 0.0055,
+      "step": 7828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023355401187088337,
+      "loss": 0.0142,
+      "step": 7829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002335379117310523,
+      "loss": 0.0131,
+      "step": 7830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023352181019598042,
+      "loss": 0.0129,
+      "step": 7831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023350570726593663,
+      "loss": 0.0206,
+      "step": 7832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023348960294118995,
+      "loss": 0.0428,
+      "step": 7833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023347349722200924,
+      "loss": 0.0075,
+      "step": 7834
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002334573901086636,
+      "loss": 0.008,
+      "step": 7835
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000233441281601422,
+      "loss": 0.0085,
+      "step": 7836
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023342517170055346,
+      "loss": 0.0222,
+      "step": 7837
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023340906040632712,
+      "loss": 0.008,
+      "step": 7838
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000233392947719012,
+      "loss": 0.0231,
+      "step": 7839
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023337683363887729,
+      "loss": 0.0097,
+      "step": 7840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023336071816619202,
+      "loss": 0.0075,
+      "step": 7841
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023334460130122536,
+      "loss": 0.0084,
+      "step": 7842
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002333284830442466,
+      "loss": 0.0287,
+      "step": 7843
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023331236339552489,
+      "loss": 0.0101,
+      "step": 7844
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002332962423553294,
+      "loss": 0.0091,
+      "step": 7845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023328011992392945,
+      "loss": 0.0118,
+      "step": 7846
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002332639961015943,
+      "loss": 0.0109,
+      "step": 7847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002332478708885932,
+      "loss": 0.0192,
+      "step": 7848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023323174428519563,
+      "loss": 0.0058,
+      "step": 7849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023321561629167074,
+      "loss": 0.0134,
+      "step": 7850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023319948690828795,
+      "loss": 0.0118,
+      "step": 7851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023318335613531675,
+      "loss": 0.0038,
+      "step": 7852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023316722397302638,
+      "loss": 0.0164,
+      "step": 7853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002331510904216865,
+      "loss": 0.0108,
+      "step": 7854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023313495548156637,
+      "loss": 0.0028,
+      "step": 7855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023311881915293555,
+      "loss": 0.0049,
+      "step": 7856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023310268143606361,
+      "loss": 0.0087,
+      "step": 7857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023308654233122,
+      "loss": 0.0099,
+      "step": 7858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002330704018386743,
+      "loss": 0.0188,
+      "step": 7859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023305425995869604,
+      "loss": 0.026,
+      "step": 7860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002330381166915549,
+      "loss": 0.002,
+      "step": 7861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023302197203752047,
+      "loss": 0.0063,
+      "step": 7862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023300582599686238,
+      "loss": 0.0163,
+      "step": 7863
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023298967856985034,
+      "loss": 0.0146,
+      "step": 7864
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000232973529756754,
+      "loss": 0.0197,
+      "step": 7865
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023295737955784303,
+      "loss": 0.0118,
+      "step": 7866
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023294122797338728,
+      "loss": 0.0229,
+      "step": 7867
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023292507500365648,
+      "loss": 0.0083,
+      "step": 7868
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023290892064892037,
+      "loss": 0.0027,
+      "step": 7869
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023289276490944877,
+      "loss": 0.0182,
+      "step": 7870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023287660778551153,
+      "loss": 0.0201,
+      "step": 7871
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023286044927737855,
+      "loss": 0.0062,
+      "step": 7872
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002328442893853196,
+      "loss": 0.002,
+      "step": 7873
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023282812810960463,
+      "loss": 0.0491,
+      "step": 7874
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002328119654505036,
+      "loss": 0.0031,
+      "step": 7875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002327958014082864,
+      "loss": 0.0033,
+      "step": 7876
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002327796359832231,
+      "loss": 0.0042,
+      "step": 7877
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023276346917558356,
+      "loss": 0.0162,
+      "step": 7878
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023274730098563786,
+      "loss": 0.0054,
+      "step": 7879
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023273113141365604,
+      "loss": 0.0122,
+      "step": 7880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023271496045990816,
+      "loss": 0.0117,
+      "step": 7881
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023269878812466434,
+      "loss": 0.0181,
+      "step": 7882
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023268261440819463,
+      "loss": 0.0037,
+      "step": 7883
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023266643931076916,
+      "loss": 0.0128,
+      "step": 7884
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023265026283265813,
+      "loss": 0.014,
+      "step": 7885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023263408497413172,
+      "loss": 0.0088,
+      "step": 7886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002326179057354601,
+      "loss": 0.0164,
+      "step": 7887
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023260172511691346,
+      "loss": 0.0132,
+      "step": 7888
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023258554311876215,
+      "loss": 0.0111,
+      "step": 7889
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002325693597412763,
+      "loss": 0.0173,
+      "step": 7890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023255317498472634,
+      "loss": 0.0153,
+      "step": 7891
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002325369888493825,
+      "loss": 0.0089,
+      "step": 7892
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023252080133551517,
+      "loss": 0.0472,
+      "step": 7893
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023250461244339469,
+      "loss": 0.0299,
+      "step": 7894
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023248842217329143,
+      "loss": 0.0202,
+      "step": 7895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002324722305254758,
+      "loss": 0.0311,
+      "step": 7896
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023245603750021826,
+      "loss": 0.0078,
+      "step": 7897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023243984309778926,
+      "loss": 0.0204,
+      "step": 7898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023242364731845922,
+      "loss": 0.0107,
+      "step": 7899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023240745016249876,
+      "loss": 0.0095,
+      "step": 7900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023239125163017826,
+      "loss": 0.0118,
+      "step": 7901
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023237505172176833,
+      "loss": 0.0324,
+      "step": 7902
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002323588504375396,
+      "loss": 0.021,
+      "step": 7903
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023234264777776255,
+      "loss": 0.0066,
+      "step": 7904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023232644374270785,
+      "loss": 0.0084,
+      "step": 7905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023231023833264618,
+      "loss": 0.0193,
+      "step": 7906
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023229403154784812,
+      "loss": 0.0165,
+      "step": 7907
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023227782338858442,
+      "loss": 0.0215,
+      "step": 7908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023226161385512574,
+      "loss": 0.01,
+      "step": 7909
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023224540294774286,
+      "loss": 0.0126,
+      "step": 7910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023222919066670647,
+      "loss": 0.016,
+      "step": 7911
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023221297701228735,
+      "loss": 0.0185,
+      "step": 7912
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023219676198475638,
+      "loss": 0.0094,
+      "step": 7913
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023218054558438433,
+      "loss": 0.0088,
+      "step": 7914
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000232164327811442,
+      "loss": 0.0089,
+      "step": 7915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023214810866620028,
+      "loss": 0.0157,
+      "step": 7916
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023213188814893014,
+      "loss": 0.0053,
+      "step": 7917
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023211566625990242,
+      "loss": 0.0128,
+      "step": 7918
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023209944299938806,
+      "loss": 0.0131,
+      "step": 7919
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023208321836765803,
+      "loss": 0.0416,
+      "step": 7920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002320669923649833,
+      "loss": 0.0127,
+      "step": 7921
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002320507649916349,
+      "loss": 0.0077,
+      "step": 7922
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023203453624788387,
+      "loss": 0.0069,
+      "step": 7923
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023201830613400117,
+      "loss": 0.0184,
+      "step": 7924
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023200207465025796,
+      "loss": 0.0398,
+      "step": 7925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002319858417969253,
+      "loss": 0.0598,
+      "step": 7926
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023196960757427437,
+      "loss": 0.0092,
+      "step": 7927
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023195337198257625,
+      "loss": 0.027,
+      "step": 7928
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002319371350221021,
+      "loss": 0.0121,
+      "step": 7929
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023192089669312312,
+      "loss": 0.0207,
+      "step": 7930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023190465699591056,
+      "loss": 0.0129,
+      "step": 7931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002318884159307356,
+      "loss": 0.0121,
+      "step": 7932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023187217349786952,
+      "loss": 0.0387,
+      "step": 7933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023185592969758362,
+      "loss": 0.0134,
+      "step": 7934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023183968453014916,
+      "loss": 0.0054,
+      "step": 7935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023182343799583744,
+      "loss": 0.019,
+      "step": 7936
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023180719009491995,
+      "loss": 0.0216,
+      "step": 7937
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023179094082766792,
+      "loss": 0.0347,
+      "step": 7938
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023177469019435275,
+      "loss": 0.0085,
+      "step": 7939
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023175843819524592,
+      "loss": 0.0202,
+      "step": 7940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023174218483061886,
+      "loss": 0.0288,
+      "step": 7941
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023172593010074304,
+      "loss": 0.0339,
+      "step": 7942
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023170967400588984,
+      "loss": 0.0086,
+      "step": 7943
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002316934165463309,
+      "loss": 0.0032,
+      "step": 7944
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023167715772233767,
+      "loss": 0.0228,
+      "step": 7945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023166089753418176,
+      "loss": 0.0035,
+      "step": 7946
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002316446359821347,
+      "loss": 0.0048,
+      "step": 7947
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002316283730664681,
+      "loss": 0.0136,
+      "step": 7948
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023161210878745356,
+      "loss": 0.0104,
+      "step": 7949
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023159584314536274,
+      "loss": 0.0156,
+      "step": 7950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023157957614046738,
+      "loss": 0.0034,
+      "step": 7951
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023156330777303907,
+      "loss": 0.0193,
+      "step": 7952
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023154703804334958,
+      "loss": 0.033,
+      "step": 7953
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002315307669516706,
+      "loss": 0.0091,
+      "step": 7954
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023151449449827392,
+      "loss": 0.0139,
+      "step": 7955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002314982206834313,
+      "loss": 0.0067,
+      "step": 7956
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023148194550741457,
+      "loss": 0.0132,
+      "step": 7957
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023146566897049557,
+      "loss": 0.0098,
+      "step": 7958
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023144939107294608,
+      "loss": 0.0295,
+      "step": 7959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023143311181503804,
+      "loss": 0.0222,
+      "step": 7960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023141683119704332,
+      "loss": 0.0161,
+      "step": 7961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023140054921923377,
+      "loss": 0.0091,
+      "step": 7962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023138426588188146,
+      "loss": 0.0156,
+      "step": 7963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023136798118525824,
+      "loss": 0.022,
+      "step": 7964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002313516951296362,
+      "loss": 0.014,
+      "step": 7965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023133540771528725,
+      "loss": 0.0059,
+      "step": 7966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023131911894248348,
+      "loss": 0.0171,
+      "step": 7967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023130282881149685,
+      "loss": 0.0139,
+      "step": 7968
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023128653732259962,
+      "loss": 0.0184,
+      "step": 7969
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023127024447606371,
+      "loss": 0.0158,
+      "step": 7970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023125395027216134,
+      "loss": 0.0508,
+      "step": 7971
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002312376547111646,
+      "loss": 0.0203,
+      "step": 7972
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023122135779334564,
+      "loss": 0.0104,
+      "step": 7973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023120505951897678,
+      "loss": 0.0307,
+      "step": 7974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023118875988833009,
+      "loss": 0.0129,
+      "step": 7975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023117245890167784,
+      "loss": 0.0105,
+      "step": 7976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023115615655929233,
+      "loss": 0.0077,
+      "step": 7977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023113985286144582,
+      "loss": 0.0067,
+      "step": 7978
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002311235478084106,
+      "loss": 0.0012,
+      "step": 7979
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023110724140045898,
+      "loss": 0.019,
+      "step": 7980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023109093363786332,
+      "loss": 0.0125,
+      "step": 7981
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000231074624520896,
+      "loss": 0.015,
+      "step": 7982
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023105831404982943,
+      "loss": 0.0073,
+      "step": 7983
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023104200222493602,
+      "loss": 0.0041,
+      "step": 7984
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023102568904648818,
+      "loss": 0.008,
+      "step": 7985
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023100937451475838,
+      "loss": 0.0044,
+      "step": 7986
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002309930586300191,
+      "loss": 0.0298,
+      "step": 7987
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023097674139254293,
+      "loss": 0.0042,
+      "step": 7988
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023096042280260225,
+      "loss": 0.0173,
+      "step": 7989
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002309441028604697,
+      "loss": 0.0251,
+      "step": 7990
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023092778156641783,
+      "loss": 0.0273,
+      "step": 7991
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023091145892071923,
+      "loss": 0.0126,
+      "step": 7992
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002308951349236466,
+      "loss": 0.0063,
+      "step": 7993
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023087880957547245,
+      "loss": 0.008,
+      "step": 7994
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023086248287646955,
+      "loss": 0.0273,
+      "step": 7995
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023084615482691053,
+      "loss": 0.0028,
+      "step": 7996
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023082982542706816,
+      "loss": 0.0201,
+      "step": 7997
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023081349467721508,
+      "loss": 0.0315,
+      "step": 7998
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023079716257762409,
+      "loss": 0.0136,
+      "step": 7999
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023078082912856798,
+      "loss": 0.0214,
+      "step": 8000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023076449433031952,
+      "loss": 0.0248,
+      "step": 8001
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023074815818315158,
+      "loss": 0.0035,
+      "step": 8002
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000230731820687337,
+      "loss": 0.0081,
+      "step": 8003
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023071548184314857,
+      "loss": 0.0163,
+      "step": 8004
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002306991416508592,
+      "loss": 0.0082,
+      "step": 8005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023068280011074197,
+      "loss": 0.0172,
+      "step": 8006
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023066645722306952,
+      "loss": 0.0306,
+      "step": 8007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002306501129881151,
+      "loss": 0.0225,
+      "step": 8008
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023063376740615148,
+      "loss": 0.0049,
+      "step": 8009
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023061742047745173,
+      "loss": 0.0388,
+      "step": 8010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023060107220228892,
+      "loss": 0.0119,
+      "step": 8011
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023058472258093602,
+      "loss": 0.011,
+      "step": 8012
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023056837161366617,
+      "loss": 0.0131,
+      "step": 8013
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002305520193007524,
+      "loss": 0.0093,
+      "step": 8014
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023053566564246788,
+      "loss": 0.0144,
+      "step": 8015
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023051931063908576,
+      "loss": 0.0345,
+      "step": 8016
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023050295429087912,
+      "loss": 0.016,
+      "step": 8017
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023048659659812118,
+      "loss": 0.0205,
+      "step": 8018
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023047023756108512,
+      "loss": 0.0053,
+      "step": 8019
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023045387718004427,
+      "loss": 0.0167,
+      "step": 8020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023043751545527173,
+      "loss": 0.014,
+      "step": 8021
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023042115238704083,
+      "loss": 0.0163,
+      "step": 8022
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023040478797562496,
+      "loss": 0.004,
+      "step": 8023
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002303884222212973,
+      "loss": 0.0336,
+      "step": 8024
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002303720551243313,
+      "loss": 0.0117,
+      "step": 8025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023035568668500022,
+      "loss": 0.0123,
+      "step": 8026
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002303393169035775,
+      "loss": 0.01,
+      "step": 8027
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023032294578033652,
+      "loss": 0.0117,
+      "step": 8028
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023030657331555074,
+      "loss": 0.0544,
+      "step": 8029
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023029019950949366,
+      "loss": 0.0078,
+      "step": 8030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002302738243624386,
+      "loss": 0.0138,
+      "step": 8031
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023025744787465917,
+      "loss": 0.0068,
+      "step": 8032
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023024107004642887,
+      "loss": 0.0163,
+      "step": 8033
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023022469087802123,
+      "loss": 0.0055,
+      "step": 8034
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023020831036970983,
+      "loss": 0.0114,
+      "step": 8035
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023019192852176821,
+      "loss": 0.0285,
+      "step": 8036
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023017554533447002,
+      "loss": 0.0026,
+      "step": 8037
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023015916080808892,
+      "loss": 0.0226,
+      "step": 8038
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002301427749428985,
+      "loss": 0.0069,
+      "step": 8039
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002301263877391725,
+      "loss": 0.0091,
+      "step": 8040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002301099991971845,
+      "loss": 0.0085,
+      "step": 8041
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023009360931720832,
+      "loss": 0.0226,
+      "step": 8042
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023007721809951767,
+      "loss": 0.0382,
+      "step": 8043
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023006082554438633,
+      "loss": 0.0383,
+      "step": 8044
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002300444316520881,
+      "loss": 0.0113,
+      "step": 8045
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023002803642289675,
+      "loss": 0.0111,
+      "step": 8046
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023001163985708612,
+      "loss": 0.019,
+      "step": 8047
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022999524195493009,
+      "loss": 0.0432,
+      "step": 8048
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002299788427167025,
+      "loss": 0.0034,
+      "step": 8049
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002299624421426773,
+      "loss": 0.0287,
+      "step": 8050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022994604023312835,
+      "loss": 0.0132,
+      "step": 8051
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022992963698832957,
+      "loss": 0.0131,
+      "step": 8052
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022991323240855507,
+      "loss": 0.0108,
+      "step": 8053
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022989682649407866,
+      "loss": 0.0093,
+      "step": 8054
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022988041924517448,
+      "loss": 0.0108,
+      "step": 8055
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002298640106621165,
+      "loss": 0.0068,
+      "step": 8056
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002298476007451788,
+      "loss": 0.0141,
+      "step": 8057
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022983118949463543,
+      "loss": 0.0285,
+      "step": 8058
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002298147769107605,
+      "loss": 0.0078,
+      "step": 8059
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022979836299382814,
+      "loss": 0.0167,
+      "step": 8060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002297819477441125,
+      "loss": 0.0022,
+      "step": 8061
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022976553116188774,
+      "loss": 0.0142,
+      "step": 8062
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000229749113247428,
+      "loss": 0.0092,
+      "step": 8063
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022973269400100753,
+      "loss": 0.0181,
+      "step": 8064
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002297162734229006,
+      "loss": 0.0366,
+      "step": 8065
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022969985151338138,
+      "loss": 0.004,
+      "step": 8066
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002296834282727243,
+      "loss": 0.0075,
+      "step": 8067
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022966700370120346,
+      "loss": 0.0073,
+      "step": 8068
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002296505777990933,
+      "loss": 0.0151,
+      "step": 8069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022963415056666817,
+      "loss": 0.0132,
+      "step": 8070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002296177220042024,
+      "loss": 0.0183,
+      "step": 8071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022960129211197037,
+      "loss": 0.0245,
+      "step": 8072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022958486089024645,
+      "loss": 0.0145,
+      "step": 8073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022956842833930514,
+      "loss": 0.0134,
+      "step": 8074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022955199445942094,
+      "loss": 0.0115,
+      "step": 8075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022953555925086828,
+      "loss": 0.0076,
+      "step": 8076
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022951912271392157,
+      "loss": 0.0107,
+      "step": 8077
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022950268484885547,
+      "loss": 0.0095,
+      "step": 8078
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022948624565594444,
+      "loss": 0.0128,
+      "step": 8079
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022946980513546304,
+      "loss": 0.006,
+      "step": 8080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022945336328768592,
+      "loss": 0.0083,
+      "step": 8081
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022943692011288765,
+      "loss": 0.0042,
+      "step": 8082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022942047561134288,
+      "loss": 0.0137,
+      "step": 8083
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002294040297833262,
+      "loss": 0.0042,
+      "step": 8084
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022938758262911236,
+      "loss": 0.0353,
+      "step": 8085
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022937113414897602,
+      "loss": 0.0097,
+      "step": 8086
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022935468434319198,
+      "loss": 0.0028,
+      "step": 8087
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022933823321203483,
+      "loss": 0.0088,
+      "step": 8088
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022932178075577947,
+      "loss": 0.0079,
+      "step": 8089
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022930532697470067,
+      "loss": 0.0024,
+      "step": 8090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022928887186907316,
+      "loss": 0.0143,
+      "step": 8091
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022927241543917184,
+      "loss": 0.0164,
+      "step": 8092
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022925595768527157,
+      "loss": 0.0057,
+      "step": 8093
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022923949860764717,
+      "loss": 0.0048,
+      "step": 8094
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022922303820657356,
+      "loss": 0.0494,
+      "step": 8095
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002292065764823257,
+      "loss": 0.0104,
+      "step": 8096
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022919011343517848,
+      "loss": 0.0109,
+      "step": 8097
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022917364906540689,
+      "loss": 0.0308,
+      "step": 8098
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022915718337328592,
+      "loss": 0.0038,
+      "step": 8099
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022914071635909058,
+      "loss": 0.0083,
+      "step": 8100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002291242480230958,
+      "loss": 0.0407,
+      "step": 8101
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022910777836557679,
+      "loss": 0.005,
+      "step": 8102
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022909130738680853,
+      "loss": 0.0321,
+      "step": 8103
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022907483508706618,
+      "loss": 0.0125,
+      "step": 8104
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002290583614666248,
+      "loss": 0.0301,
+      "step": 8105
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022904188652575955,
+      "loss": 0.0838,
+      "step": 8106
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022902541026474557,
+      "loss": 0.0174,
+      "step": 8107
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022900893268385812,
+      "loss": 0.0675,
+      "step": 8108
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022899245378337233,
+      "loss": 0.0101,
+      "step": 8109
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022897597356356345,
+      "loss": 0.0063,
+      "step": 8110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022895949202470672,
+      "loss": 0.0089,
+      "step": 8111
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022894300916707742,
+      "loss": 0.0093,
+      "step": 8112
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002289265249909509,
+      "loss": 0.0123,
+      "step": 8113
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022891003949660242,
+      "loss": 0.0034,
+      "step": 8114
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002288935526843073,
+      "loss": 0.029,
+      "step": 8115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022887706455434087,
+      "loss": 0.0369,
+      "step": 8116
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022886057510697867,
+      "loss": 0.0051,
+      "step": 8117
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022884408434249598,
+      "loss": 0.0061,
+      "step": 8118
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022882759226116825,
+      "loss": 0.0141,
+      "step": 8119
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022881109886327086,
+      "loss": 0.017,
+      "step": 8120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022879460414907944,
+      "loss": 0.0066,
+      "step": 8121
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022877810811886942,
+      "loss": 0.0101,
+      "step": 8122
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022876161077291623,
+      "loss": 0.0064,
+      "step": 8123
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022874511211149547,
+      "loss": 0.0103,
+      "step": 8124
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022872861213488267,
+      "loss": 0.0236,
+      "step": 8125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022871211084335349,
+      "loss": 0.0044,
+      "step": 8126
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022869560823718347,
+      "loss": 0.018,
+      "step": 8127
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022867910431664823,
+      "loss": 0.0327,
+      "step": 8128
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002286625990820234,
+      "loss": 0.0209,
+      "step": 8129
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022864609253358474,
+      "loss": 0.0093,
+      "step": 8130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022862958467160786,
+      "loss": 0.008,
+      "step": 8131
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002286130754963685,
+      "loss": 0.0066,
+      "step": 8132
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022859656500814237,
+      "loss": 0.0259,
+      "step": 8133
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022858005320720523,
+      "loss": 0.0049,
+      "step": 8134
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022856354009383287,
+      "loss": 0.0053,
+      "step": 8135
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022854702566830112,
+      "loss": 0.0146,
+      "step": 8136
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022853050993088575,
+      "loss": 0.0144,
+      "step": 8137
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022851399288186265,
+      "loss": 0.0065,
+      "step": 8138
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022849747452150762,
+      "loss": 0.0171,
+      "step": 8139
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022848095485009665,
+      "loss": 0.0052,
+      "step": 8140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022846443386790555,
+      "loss": 0.0127,
+      "step": 8141
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022844791157521028,
+      "loss": 0.0096,
+      "step": 8142
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022843138797228686,
+      "loss": 0.0025,
+      "step": 8143
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022841486305941118,
+      "loss": 0.0173,
+      "step": 8144
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022839833683685929,
+      "loss": 0.0093,
+      "step": 8145
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022838180930490713,
+      "loss": 0.0103,
+      "step": 8146
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022836528046383086,
+      "loss": 0.0252,
+      "step": 8147
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022834875031390643,
+      "loss": 0.0087,
+      "step": 8148
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022833221885541,
+      "loss": 0.0153,
+      "step": 8149
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002283156860886177,
+      "loss": 0.0157,
+      "step": 8150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022829915201380555,
+      "loss": 0.0171,
+      "step": 8151
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022828261663124976,
+      "loss": 0.0116,
+      "step": 8152
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022826607994122654,
+      "loss": 0.0169,
+      "step": 8153
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022824954194401205,
+      "loss": 0.024,
+      "step": 8154
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002282330026398825,
+      "loss": 0.0114,
+      "step": 8155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002282164620291141,
+      "loss": 0.0097,
+      "step": 8156
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022819992011198308,
+      "loss": 0.0458,
+      "step": 8157
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022818337688876588,
+      "loss": 0.0196,
+      "step": 8158
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002281668323597387,
+      "loss": 0.0105,
+      "step": 8159
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022815028652517784,
+      "loss": 0.021,
+      "step": 8160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022813373938535968,
+      "loss": 0.0071,
+      "step": 8161
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022811719094056057,
+      "loss": 0.0207,
+      "step": 8162
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022810064119105694,
+      "loss": 0.0228,
+      "step": 8163
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002280840901371252,
+      "loss": 0.0063,
+      "step": 8164
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002280675377790417,
+      "loss": 0.0145,
+      "step": 8165
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000228050984117083,
+      "loss": 0.005,
+      "step": 8166
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022803442915152552,
+      "loss": 0.022,
+      "step": 8167
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022801787288264578,
+      "loss": 0.0059,
+      "step": 8168
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022800131531072024,
+      "loss": 0.0036,
+      "step": 8169
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022798475643602556,
+      "loss": 0.0159,
+      "step": 8170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022796819625883822,
+      "loss": 0.0319,
+      "step": 8171
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022795163477943485,
+      "loss": 0.0055,
+      "step": 8172
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022793507199809205,
+      "loss": 0.013,
+      "step": 8173
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022791850791508637,
+      "loss": 0.0028,
+      "step": 8174
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022790194253069455,
+      "loss": 0.0243,
+      "step": 8175
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022788537584519326,
+      "loss": 0.0305,
+      "step": 8176
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022786880785885915,
+      "loss": 0.0143,
+      "step": 8177
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000227852238571969,
+      "loss": 0.0071,
+      "step": 8178
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022783566798479944,
+      "loss": 0.0074,
+      "step": 8179
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022781909609762731,
+      "loss": 0.0092,
+      "step": 8180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022780252291072943,
+      "loss": 0.0154,
+      "step": 8181
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022778594842438256,
+      "loss": 0.0028,
+      "step": 8182
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002277693726388635,
+      "loss": 0.0182,
+      "step": 8183
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002277527955544491,
+      "loss": 0.0232,
+      "step": 8184
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022773621717141627,
+      "loss": 0.0048,
+      "step": 8185
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022771963749004185,
+      "loss": 0.037,
+      "step": 8186
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002277030565106028,
+      "loss": 0.0038,
+      "step": 8187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000227686474233376,
+      "loss": 0.0063,
+      "step": 8188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022766989065863845,
+      "loss": 0.0088,
+      "step": 8189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022765330578666717,
+      "loss": 0.02,
+      "step": 8190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022763671961773907,
+      "loss": 0.0122,
+      "step": 8191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022762013215213116,
+      "loss": 0.0127,
+      "step": 8192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022760354339012055,
+      "loss": 0.0149,
+      "step": 8193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002275869533319843,
+      "loss": 0.0057,
+      "step": 8194
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022757036197799943,
+      "loss": 0.0157,
+      "step": 8195
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022755376932844315,
+      "loss": 0.0063,
+      "step": 8196
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022753717538359246,
+      "loss": 0.0182,
+      "step": 8197
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022752058014372464,
+      "loss": 0.0128,
+      "step": 8198
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022750398360911676,
+      "loss": 0.0391,
+      "step": 8199
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022748738578004605,
+      "loss": 0.0099,
+      "step": 8200
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022747078665678975,
+      "loss": 0.0185,
+      "step": 8201
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002274541862396251,
+      "loss": 0.0289,
+      "step": 8202
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002274375845288293,
+      "loss": 0.0052,
+      "step": 8203
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022742098152467964,
+      "loss": 0.0106,
+      "step": 8204
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022740437722745355,
+      "loss": 0.0106,
+      "step": 8205
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022738777163742815,
+      "loss": 0.0068,
+      "step": 8206
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022737116475488094,
+      "loss": 0.0064,
+      "step": 8207
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022735455658008915,
+      "loss": 0.0185,
+      "step": 8208
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022733794711333034,
+      "loss": 0.0047,
+      "step": 8209
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022732133635488177,
+      "loss": 0.0181,
+      "step": 8210
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022730472430502094,
+      "loss": 0.0039,
+      "step": 8211
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002272881109640253,
+      "loss": 0.0184,
+      "step": 8212
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002272714963321723,
+      "loss": 0.0145,
+      "step": 8213
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022725488040973947,
+      "loss": 0.0109,
+      "step": 8214
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002272382631970043,
+      "loss": 0.0084,
+      "step": 8215
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022722164469424436,
+      "loss": 0.0069,
+      "step": 8216
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022720502490173718,
+      "loss": 0.0104,
+      "step": 8217
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002271884038197603,
+      "loss": 0.0197,
+      "step": 8218
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022717178144859146,
+      "loss": 0.0264,
+      "step": 8219
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002271551577885081,
+      "loss": 0.0068,
+      "step": 8220
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022713853283978802,
+      "loss": 0.0103,
+      "step": 8221
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022712190660270884,
+      "loss": 0.0238,
+      "step": 8222
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022710527907754822,
+      "loss": 0.0206,
+      "step": 8223
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022708865026458392,
+      "loss": 0.0028,
+      "step": 8224
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022707202016409363,
+      "loss": 0.0137,
+      "step": 8225
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002270553887763551,
+      "loss": 0.0058,
+      "step": 8226
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022703875610164613,
+      "loss": 0.0056,
+      "step": 8227
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022702212214024454,
+      "loss": 0.0145,
+      "step": 8228
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002270054868924281,
+      "loss": 0.0252,
+      "step": 8229
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022698885035847465,
+      "loss": 0.0125,
+      "step": 8230
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002269722125386621,
+      "loss": 0.021,
+      "step": 8231
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022695557343326827,
+      "loss": 0.0096,
+      "step": 8232
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022693893304257117,
+      "loss": 0.0101,
+      "step": 8233
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022692229136684862,
+      "loss": 0.0131,
+      "step": 8234
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002269056484063786,
+      "loss": 0.0119,
+      "step": 8235
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022688900416143905,
+      "loss": 0.0064,
+      "step": 8236
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022687235863230805,
+      "loss": 0.0207,
+      "step": 8237
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022685571181926354,
+      "loss": 0.009,
+      "step": 8238
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022683906372258355,
+      "loss": 0.0107,
+      "step": 8239
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022682241434254614,
+      "loss": 0.0059,
+      "step": 8240
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022680576367942938,
+      "loss": 0.0199,
+      "step": 8241
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022678911173351144,
+      "loss": 0.0161,
+      "step": 8242
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022677245850507037,
+      "loss": 0.015,
+      "step": 8243
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002267558039943843,
+      "loss": 0.006,
+      "step": 8244
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022673914820173146,
+      "loss": 0.0104,
+      "step": 8245
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022672249112738995,
+      "loss": 0.0157,
+      "step": 8246
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022670583277163809,
+      "loss": 0.0203,
+      "step": 8247
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022668917313475393,
+      "loss": 0.0045,
+      "step": 8248
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022667251221701587,
+      "loss": 0.0168,
+      "step": 8249
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022665585001870208,
+      "loss": 0.0109,
+      "step": 8250
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022663918654009093,
+      "loss": 0.0335,
+      "step": 8251
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022662252178146073,
+      "loss": 0.0207,
+      "step": 8252
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002266058557430897,
+      "loss": 0.0037,
+      "step": 8253
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022658918842525632,
+      "loss": 0.0157,
+      "step": 8254
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022657251982823894,
+      "loss": 0.0106,
+      "step": 8255
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022655584995231594,
+      "loss": 0.0211,
+      "step": 8256
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022653917879776573,
+      "loss": 0.0169,
+      "step": 8257
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022652250636486673,
+      "loss": 0.0161,
+      "step": 8258
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022650583265389744,
+      "loss": 0.0298,
+      "step": 8259
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002264891576651363,
+      "loss": 0.0256,
+      "step": 8260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002264724813988619,
+      "loss": 0.0041,
+      "step": 8261
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022645580385535267,
+      "loss": 0.0252,
+      "step": 8262
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022643912503488722,
+      "loss": 0.011,
+      "step": 8263
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002264224449377441,
+      "loss": 0.0125,
+      "step": 8264
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022640576356420186,
+      "loss": 0.005,
+      "step": 8265
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022638908091453917,
+      "loss": 0.0269,
+      "step": 8266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022637239698903463,
+      "loss": 0.0046,
+      "step": 8267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022635571178796686,
+      "loss": 0.0194,
+      "step": 8268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002263390253116146,
+      "loss": 0.0083,
+      "step": 8269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022632233756025654,
+      "loss": 0.0104,
+      "step": 8270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002263056485341713,
+      "loss": 0.0052,
+      "step": 8271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022628895823363775,
+      "loss": 0.0122,
+      "step": 8272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002262722666589346,
+      "loss": 0.0132,
+      "step": 8273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022625557381034057,
+      "loss": 0.009,
+      "step": 8274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002262388796881346,
+      "loss": 0.0067,
+      "step": 8275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002262221842925954,
+      "loss": 0.0229,
+      "step": 8276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002262054876240018,
+      "loss": 0.0068,
+      "step": 8277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022618878968263273,
+      "loss": 0.0214,
+      "step": 8278
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022617209046876707,
+      "loss": 0.0099,
+      "step": 8279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002261553899826837,
+      "loss": 0.046,
+      "step": 8280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022613868822466155,
+      "loss": 0.0289,
+      "step": 8281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002261219851949796,
+      "loss": 0.0208,
+      "step": 8282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002261052808939168,
+      "loss": 0.0125,
+      "step": 8283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022608857532175215,
+      "loss": 0.0163,
+      "step": 8284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002260718684787647,
+      "loss": 0.0057,
+      "step": 8285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002260551603652334,
+      "loss": 0.0104,
+      "step": 8286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022603845098143738,
+      "loss": 0.0114,
+      "step": 8287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022602174032765573,
+      "loss": 0.0083,
+      "step": 8288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022600502840416746,
+      "loss": 0.0187,
+      "step": 8289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022598831521125176,
+      "loss": 0.0403,
+      "step": 8290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022597160074918778,
+      "loss": 0.0094,
+      "step": 8291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022595488501825464,
+      "loss": 0.0045,
+      "step": 8292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022593816801873151,
+      "loss": 0.0071,
+      "step": 8293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002259214497508977,
+      "loss": 0.0234,
+      "step": 8294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022590473021503235,
+      "loss": 0.0278,
+      "step": 8295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022588800941141473,
+      "loss": 0.0289,
+      "step": 8296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022587128734032405,
+      "loss": 0.02,
+      "step": 8297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022585456400203973,
+      "loss": 0.0069,
+      "step": 8298
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022583783939684096,
+      "loss": 0.0089,
+      "step": 8299
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022582111352500712,
+      "loss": 0.0362,
+      "step": 8300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022580438638681756,
+      "loss": 0.0104,
+      "step": 8301
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022578765798255166,
+      "loss": 0.0053,
+      "step": 8302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022577092831248884,
+      "loss": 0.0114,
+      "step": 8303
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022575419737690843,
+      "loss": 0.0121,
+      "step": 8304
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022573746517609,
+      "loss": 0.004,
+      "step": 8305
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022572073171031286,
+      "loss": 0.0148,
+      "step": 8306
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022570399697985667,
+      "loss": 0.0025,
+      "step": 8307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022568726098500072,
+      "loss": 0.0265,
+      "step": 8308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022567052372602473,
+      "loss": 0.0129,
+      "step": 8309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002256537852032081,
+      "loss": 0.0258,
+      "step": 8310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022563704541683045,
+      "loss": 0.0033,
+      "step": 8311
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022562030436717138,
+      "loss": 0.012,
+      "step": 8312
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022560356205451047,
+      "loss": 0.0081,
+      "step": 8313
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022558681847912736,
+      "loss": 0.0106,
+      "step": 8314
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002255700736413017,
+      "loss": 0.0076,
+      "step": 8315
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022555332754131322,
+      "loss": 0.0086,
+      "step": 8316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022553658017944152,
+      "loss": 0.0089,
+      "step": 8317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022551983155596634,
+      "loss": 0.0205,
+      "step": 8318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022550308167116741,
+      "loss": 0.0064,
+      "step": 8319
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022548633052532448,
+      "loss": 0.019,
+      "step": 8320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022546957811871743,
+      "loss": 0.0148,
+      "step": 8321
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022545282445162585,
+      "loss": 0.0173,
+      "step": 8322
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022543606952432974,
+      "loss": 0.0027,
+      "step": 8323
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022541931333710881,
+      "loss": 0.0072,
+      "step": 8324
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022540255589024305,
+      "loss": 0.0131,
+      "step": 8325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022538579718401223,
+      "loss": 0.0084,
+      "step": 8326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022536903721869634,
+      "loss": 0.0124,
+      "step": 8327
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002253522759945752,
+      "loss": 0.0367,
+      "step": 8328
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002253355135119289,
+      "loss": 0.0203,
+      "step": 8329
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022531874977103726,
+      "loss": 0.0061,
+      "step": 8330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022530198477218035,
+      "loss": 0.002,
+      "step": 8331
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022528521851563813,
+      "loss": 0.0173,
+      "step": 8332
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022526845100169068,
+      "loss": 0.0185,
+      "step": 8333
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022525168223061796,
+      "loss": 0.0075,
+      "step": 8334
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022523491220270017,
+      "loss": 0.0035,
+      "step": 8335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022521814091821724,
+      "loss": 0.0143,
+      "step": 8336
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022520136837744944,
+      "loss": 0.0161,
+      "step": 8337
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251845945806768,
+      "loss": 0.0445,
+      "step": 8338
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251678195281796,
+      "loss": 0.0112,
+      "step": 8339
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251510432202378,
+      "loss": 0.0218,
+      "step": 8340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251342656571318,
+      "loss": 0.0075,
+      "step": 8341
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022511748683914172,
+      "loss": 0.0142,
+      "step": 8342
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251007067665478,
+      "loss": 0.0111,
+      "step": 8343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002250839254396303,
+      "loss": 0.0219,
+      "step": 8344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022506714285866955,
+      "loss": 0.0025,
+      "step": 8345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022505035902394582,
+      "loss": 0.0031,
+      "step": 8346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002250335739357394,
+      "loss": 0.0057,
+      "step": 8347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022501678759433063,
+      "loss": 0.0176,
+      "step": 8348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000225,
+      "loss": 0.03,
+      "step": 8349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022498321115302773,
+      "loss": 0.0104,
+      "step": 8350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002249664210536943,
+      "loss": 0.0102,
+      "step": 8351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022494962970228012,
+      "loss": 0.0022,
+      "step": 8352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002249328370990657,
+      "loss": 0.0086,
+      "step": 8353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022491604324433137,
+      "loss": 0.0095,
+      "step": 8354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022489924813835774,
+      "loss": 0.0099,
+      "step": 8355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022488245178142526,
+      "loss": 0.026,
+      "step": 8356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022486565417381455,
+      "loss": 0.015,
+      "step": 8357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002248488553158061,
+      "loss": 0.0176,
+      "step": 8358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022483205520768045,
+      "loss": 0.0043,
+      "step": 8359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002248152538497182,
+      "loss": 0.0161,
+      "step": 8360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002247984512422,
+      "loss": 0.0089,
+      "step": 8361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022478164738540642,
+      "loss": 0.0067,
+      "step": 8362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022476484227961828,
+      "loss": 0.0145,
+      "step": 8363
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022474803592511608,
+      "loss": 0.0053,
+      "step": 8364
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022473122832218054,
+      "loss": 0.0168,
+      "step": 8365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002247144194710925,
+      "loss": 0.0334,
+      "step": 8366
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002246976093721326,
+      "loss": 0.0181,
+      "step": 8367
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002246807980255816,
+      "loss": 0.0065,
+      "step": 8368
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022466398543172033,
+      "loss": 0.0039,
+      "step": 8369
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022464717159082954,
+      "loss": 0.0108,
+      "step": 8370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022463035650319008,
+      "loss": 0.0113,
+      "step": 8371
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022461354016908283,
+      "loss": 0.018,
+      "step": 8372
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002245967225887886,
+      "loss": 0.0333,
+      "step": 8373
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022457990376258826,
+      "loss": 0.0082,
+      "step": 8374
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022456308369076276,
+      "loss": 0.0172,
+      "step": 8375
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022454626237359301,
+      "loss": 0.0074,
+      "step": 8376
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022452943981136,
+      "loss": 0.005,
+      "step": 8377
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002245126160043446,
+      "loss": 0.0009,
+      "step": 8378
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002244957909528279,
+      "loss": 0.0113,
+      "step": 8379
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022447896465709088,
+      "loss": 0.0329,
+      "step": 8380
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022446213711741454,
+      "loss": 0.0134,
+      "step": 8381
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022444530833407999,
+      "loss": 0.0106,
+      "step": 8382
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002244284783073682,
+      "loss": 0.0613,
+      "step": 8383
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022441164703756037,
+      "loss": 0.0171,
+      "step": 8384
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022439481452493756,
+      "loss": 0.023,
+      "step": 8385
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022437798076978092,
+      "loss": 0.0126,
+      "step": 8386
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002243611457723716,
+      "loss": 0.0038,
+      "step": 8387
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022434430953299078,
+      "loss": 0.015,
+      "step": 8388
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022432747205191966,
+      "loss": 0.0063,
+      "step": 8389
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022431063332943948,
+      "loss": 0.0127,
+      "step": 8390
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022429379336583145,
+      "loss": 0.0088,
+      "step": 8391
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022427695216137677,
+      "loss": 0.0353,
+      "step": 8392
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022426010971635682,
+      "loss": 0.0013,
+      "step": 8393
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022424326603105287,
+      "loss": 0.0088,
+      "step": 8394
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022422642110574622,
+      "loss": 0.0133,
+      "step": 8395
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022420957494071822,
+      "loss": 0.0349,
+      "step": 8396
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022419272753625025,
+      "loss": 0.0106,
+      "step": 8397
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022417587889262368,
+      "loss": 0.0126,
+      "step": 8398
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022415902901011991,
+      "loss": 0.0138,
+      "step": 8399
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002241421778890204,
+      "loss": 0.013,
+      "step": 8400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002241253255296065,
+      "loss": 0.0298,
+      "step": 8401
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022410847193215982,
+      "loss": 0.0049,
+      "step": 8402
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002240916170969617,
+      "loss": 0.0124,
+      "step": 8403
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002240747610242938,
+      "loss": 0.0105,
+      "step": 8404
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002240579037144375,
+      "loss": 0.0182,
+      "step": 8405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022404104516767442,
+      "loss": 0.0057,
+      "step": 8406
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002240241853842861,
+      "loss": 0.0083,
+      "step": 8407
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022400732436455418,
+      "loss": 0.0175,
+      "step": 8408
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022399046210876028,
+      "loss": 0.0481,
+      "step": 8409
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022397359861718596,
+      "loss": 0.0072,
+      "step": 8410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022395673389011291,
+      "loss": 0.016,
+      "step": 8411
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002239398679278228,
+      "loss": 0.0064,
+      "step": 8412
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022392300073059736,
+      "loss": 0.0194,
+      "step": 8413
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002239061322987182,
+      "loss": 0.0078,
+      "step": 8414
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022388926263246716,
+      "loss": 0.0222,
+      "step": 8415
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000223872391732126,
+      "loss": 0.0012,
+      "step": 8416
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002238555195979764,
+      "loss": 0.0263,
+      "step": 8417
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002238386462303002,
+      "loss": 0.0226,
+      "step": 8418
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022382177162937923,
+      "loss": 0.018,
+      "step": 8419
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022380489579549534,
+      "loss": 0.0215,
+      "step": 8420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022378801872893037,
+      "loss": 0.0161,
+      "step": 8421
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022377114042996624,
+      "loss": 0.0082,
+      "step": 8422
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002237542608988848,
+      "loss": 0.0075,
+      "step": 8423
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022373738013596793,
+      "loss": 0.0038,
+      "step": 8424
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022372049814149768,
+      "loss": 0.005,
+      "step": 8425
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022370361491575594,
+      "loss": 0.0193,
+      "step": 8426
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002236867304590247,
+      "loss": 0.0076,
+      "step": 8427
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022366984477158592,
+      "loss": 0.017,
+      "step": 8428
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022365295785372176,
+      "loss": 0.0123,
+      "step": 8429
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002236360697057141,
+      "loss": 0.0112,
+      "step": 8430
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022361918032784512,
+      "loss": 0.0663,
+      "step": 8431
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022360228972039686,
+      "loss": 0.0047,
+      "step": 8432
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002235853978836514,
+      "loss": 0.0124,
+      "step": 8433
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002235685048178909,
+      "loss": 0.0046,
+      "step": 8434
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022355161052339752,
+      "loss": 0.0066,
+      "step": 8435
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022353471500045338,
+      "loss": 0.0104,
+      "step": 8436
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022351781824934072,
+      "loss": 0.0382,
+      "step": 8437
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022350092027034169,
+      "loss": 0.0212,
+      "step": 8438
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022348402106373857,
+      "loss": 0.0029,
+      "step": 8439
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022346712062981357,
+      "loss": 0.0269,
+      "step": 8440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022345021896884903,
+      "loss": 0.0222,
+      "step": 8441
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022343331608112713,
+      "loss": 0.0019,
+      "step": 8442
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022341641196693026,
+      "loss": 0.0152,
+      "step": 8443
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002233995066265407,
+      "loss": 0.0105,
+      "step": 8444
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022338260006024086,
+      "loss": 0.0084,
+      "step": 8445
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002233656922683131,
+      "loss": 0.0285,
+      "step": 8446
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022334878325103974,
+      "loss": 0.0074,
+      "step": 8447
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022333187300870323,
+      "loss": 0.012,
+      "step": 8448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022331496154158606,
+      "loss": 0.0339,
+      "step": 8449
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022329804884997065,
+      "loss": 0.0167,
+      "step": 8450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022328113493413944,
+      "loss": 0.0094,
+      "step": 8451
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022326421979437496,
+      "loss": 0.011,
+      "step": 8452
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022324730343095968,
+      "loss": 0.0042,
+      "step": 8453
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022323038584417618,
+      "loss": 0.0216,
+      "step": 8454
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022321346703430705,
+      "loss": 0.0051,
+      "step": 8455
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022319654700163482,
+      "loss": 0.0216,
+      "step": 8456
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022317962574644205,
+      "loss": 0.029,
+      "step": 8457
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002231627032690114,
+      "loss": 0.0085,
+      "step": 8458
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022314577956962547,
+      "loss": 0.0109,
+      "step": 8459
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000223128854648567,
+      "loss": 0.0133,
+      "step": 8460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022311192850611858,
+      "loss": 0.0061,
+      "step": 8461
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022309500114256296,
+      "loss": 0.0192,
+      "step": 8462
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022307807255818283,
+      "loss": 0.0115,
+      "step": 8463
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022306114275326102,
+      "loss": 0.0231,
+      "step": 8464
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022304421172808012,
+      "loss": 0.0064,
+      "step": 8465
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022302727948292304,
+      "loss": 0.0189,
+      "step": 8466
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022301034601807253,
+      "loss": 0.0329,
+      "step": 8467
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022299341133381144,
+      "loss": 0.0139,
+      "step": 8468
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002229764754304226,
+      "loss": 0.0056,
+      "step": 8469
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022295953830818885,
+      "loss": 0.0099,
+      "step": 8470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022294259996739308,
+      "loss": 0.0153,
+      "step": 8471
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002229256604083182,
+      "loss": 0.0105,
+      "step": 8472
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022290871963124719,
+      "loss": 0.0077,
+      "step": 8473
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022289177763646292,
+      "loss": 0.0077,
+      "step": 8474
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022287483442424833,
+      "loss": 0.0134,
+      "step": 8475
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002228578899948865,
+      "loss": 0.0183,
+      "step": 8476
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022284094434866033,
+      "loss": 0.0051,
+      "step": 8477
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022282399748585293,
+      "loss": 0.0225,
+      "step": 8478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022280704940674722,
+      "loss": 0.0356,
+      "step": 8479
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002227901001116264,
+      "loss": 0.0026,
+      "step": 8480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022277314960077356,
+      "loss": 0.0105,
+      "step": 8481
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002227561978744717,
+      "loss": 0.0085,
+      "step": 8482
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022273924493300404,
+      "loss": 0.0011,
+      "step": 8483
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022272229077665366,
+      "loss": 0.0041,
+      "step": 8484
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022270533540570374,
+      "loss": 0.0053,
+      "step": 8485
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022268837882043747,
+      "loss": 0.0127,
+      "step": 8486
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002226714210211381,
+      "loss": 0.0219,
+      "step": 8487
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022265446200808882,
+      "loss": 0.0171,
+      "step": 8488
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022263750178157284,
+      "loss": 0.0221,
+      "step": 8489
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002226205403418735,
+      "loss": 0.0125,
+      "step": 8490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000222603577689274,
+      "loss": 0.0192,
+      "step": 8491
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002225866138240578,
+      "loss": 0.0257,
+      "step": 8492
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022256964874650808,
+      "loss": 0.0165,
+      "step": 8493
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022255268245690825,
+      "loss": 0.0083,
+      "step": 8494
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022253571495554165,
+      "loss": 0.0323,
+      "step": 8495
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002225187462426917,
+      "loss": 0.0129,
+      "step": 8496
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002225017763186418,
+      "loss": 0.0185,
+      "step": 8497
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022248480518367538,
+      "loss": 0.0119,
+      "step": 8498
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022246783283807588,
+      "loss": 0.0122,
+      "step": 8499
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022245085928212678,
+      "loss": 0.0138,
+      "step": 8500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022243388451611158,
+      "loss": 0.0046,
+      "step": 8501
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022241690854031378,
+      "loss": 0.0018,
+      "step": 8502
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002223999313550169,
+      "loss": 0.0338,
+      "step": 8503
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022238295296050453,
+      "loss": 0.0098,
+      "step": 8504
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022236597335706016,
+      "loss": 0.016,
+      "step": 8505
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002223489925449675,
+      "loss": 0.0226,
+      "step": 8506
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022233201052451005,
+      "loss": 0.0096,
+      "step": 8507
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002223150272959715,
+      "loss": 0.0051,
+      "step": 8508
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002222980428596355,
+      "loss": 0.0101,
+      "step": 8509
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002222810572157857,
+      "loss": 0.0088,
+      "step": 8510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022226407036470577,
+      "loss": 0.0206,
+      "step": 8511
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002222470823066795,
+      "loss": 0.0246,
+      "step": 8512
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002222300930419906,
+      "loss": 0.0072,
+      "step": 8513
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022221310257092272,
+      "loss": 0.0281,
+      "step": 8514
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022219611089375983,
+      "loss": 0.008,
+      "step": 8515
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022217911801078553,
+      "loss": 0.0142,
+      "step": 8516
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022216212392228374,
+      "loss": 0.0068,
+      "step": 8517
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022214512862853825,
+      "loss": 0.0096,
+      "step": 8518
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022212813212983294,
+      "loss": 0.0068,
+      "step": 8519
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022211113442645165,
+      "loss": 0.0169,
+      "step": 8520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002220941355186783,
+      "loss": 0.0378,
+      "step": 8521
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002220771354067968,
+      "loss": 0.0223,
+      "step": 8522
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002220601340910911,
+      "loss": 0.0061,
+      "step": 8523
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022204313157184514,
+      "loss": 0.0245,
+      "step": 8524
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002220261278493429,
+      "loss": 0.0127,
+      "step": 8525
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022200912292386833,
+      "loss": 0.0102,
+      "step": 8526
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002219921167957055,
+      "loss": 0.0311,
+      "step": 8527
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022197510946513844,
+      "loss": 0.0227,
+      "step": 8528
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022195810093245116,
+      "loss": 0.0015,
+      "step": 8529
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002219410911979278,
+      "loss": 0.0165,
+      "step": 8530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022192408026185237,
+      "loss": 0.0238,
+      "step": 8531
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022190706812450905,
+      "loss": 0.0276,
+      "step": 8532
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022189005478618197,
+      "loss": 0.0325,
+      "step": 8533
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022187304024715528,
+      "loss": 0.0064,
+      "step": 8534
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022185602450771317,
+      "loss": 0.0133,
+      "step": 8535
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022183900756813973,
+      "loss": 0.0067,
+      "step": 8536
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022182198942871932,
+      "loss": 0.0512,
+      "step": 8537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022180497008973614,
+      "loss": 0.0142,
+      "step": 8538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022178794955147435,
+      "loss": 0.0129,
+      "step": 8539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022177092781421832,
+      "loss": 0.0183,
+      "step": 8540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022175390487825231,
+      "loss": 0.0053,
+      "step": 8541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002217368807438607,
+      "loss": 0.0049,
+      "step": 8542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002217198554113277,
+      "loss": 0.043,
+      "step": 8543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022170282888093774,
+      "loss": 0.0143,
+      "step": 8544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002216858011529752,
+      "loss": 0.0053,
+      "step": 8545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022166877222772448,
+      "loss": 0.0085,
+      "step": 8546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022165174210546997,
+      "loss": 0.0208,
+      "step": 8547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022163471078649613,
+      "loss": 0.006,
+      "step": 8548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022161767827108737,
+      "loss": 0.0145,
+      "step": 8549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002216006445595282,
+      "loss": 0.0146,
+      "step": 8550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002215836096521031,
+      "loss": 0.0348,
+      "step": 8551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022156657354909664,
+      "loss": 0.0168,
+      "step": 8552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022154953625079328,
+      "loss": 0.0155,
+      "step": 8553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002215324977574776,
+      "loss": 0.0191,
+      "step": 8554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022151545806943417,
+      "loss": 0.0058,
+      "step": 8555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022149841718694765,
+      "loss": 0.0305,
+      "step": 8556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002214813751103026,
+      "loss": 0.0165,
+      "step": 8557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002214643318397836,
+      "loss": 0.0152,
+      "step": 8558
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022144728737567536,
+      "loss": 0.0242,
+      "step": 8559
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002214302417182626,
+      "loss": 0.0179,
+      "step": 8560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022141319486782994,
+      "loss": 0.0052,
+      "step": 8561
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002213961468246621,
+      "loss": 0.0106,
+      "step": 8562
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022137909758904384,
+      "loss": 0.0043,
+      "step": 8563
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022136204716125996,
+      "loss": 0.0192,
+      "step": 8564
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022134499554159518,
+      "loss": 0.0206,
+      "step": 8565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022132794273033428,
+      "loss": 0.0361,
+      "step": 8566
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002213108887277621,
+      "loss": 0.0365,
+      "step": 8567
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022129383353416347,
+      "loss": 0.014,
+      "step": 8568
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022127677714982327,
+      "loss": 0.0202,
+      "step": 8569
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002212597195750263,
+      "loss": 0.0079,
+      "step": 8570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022124266081005755,
+      "loss": 0.0108,
+      "step": 8571
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022122560085520185,
+      "loss": 0.0265,
+      "step": 8572
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022120853971074414,
+      "loss": 0.0176,
+      "step": 8573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022119147737696942,
+      "loss": 0.0079,
+      "step": 8574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002211744138541627,
+      "loss": 0.0051,
+      "step": 8575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022115734914260888,
+      "loss": 0.0114,
+      "step": 8576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000221140283242593,
+      "loss": 0.0024,
+      "step": 8577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022112321615440011,
+      "loss": 0.0422,
+      "step": 8578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022110614787831525,
+      "loss": 0.0249,
+      "step": 8579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022108907841462353,
+      "loss": 0.0329,
+      "step": 8580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022107200776360999,
+      "loss": 0.0032,
+      "step": 8581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022105493592555975,
+      "loss": 0.009,
+      "step": 8582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022103786290075795,
+      "loss": 0.0086,
+      "step": 8583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022102078868948977,
+      "loss": 0.01,
+      "step": 8584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022100371329204034,
+      "loss": 0.0127,
+      "step": 8585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022098663670869486,
+      "loss": 0.0088,
+      "step": 8586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022096955893973857,
+      "loss": 0.0024,
+      "step": 8587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002209524799854567,
+      "loss": 0.0108,
+      "step": 8588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022093539984613446,
+      "loss": 0.019,
+      "step": 8589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022091831852205716,
+      "loss": 0.0175,
+      "step": 8590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022090123601351,
+      "loss": 0.0059,
+      "step": 8591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022088415232077846,
+      "loss": 0.0087,
+      "step": 8592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022086706744414768,
+      "loss": 0.006,
+      "step": 8593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022084998138390316,
+      "loss": 0.0177,
+      "step": 8594
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002208328941403302,
+      "loss": 0.0128,
+      "step": 8595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022081580571371422,
+      "loss": 0.0149,
+      "step": 8596
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002207987161043406,
+      "loss": 0.0304,
+      "step": 8597
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022078162531249476,
+      "loss": 0.0204,
+      "step": 8598
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002207645333384622,
+      "loss": 0.0365,
+      "step": 8599
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022074744018252832,
+      "loss": 0.0099,
+      "step": 8600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022073034584497867,
+      "loss": 0.012,
+      "step": 8601
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022071325032609873,
+      "loss": 0.0138,
+      "step": 8602
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000220696153626174,
+      "loss": 0.0118,
+      "step": 8603
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022067905574549008,
+      "loss": 0.0141,
+      "step": 8604
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002206619566843325,
+      "loss": 0.0098,
+      "step": 8605
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022064485644298688,
+      "loss": 0.0107,
+      "step": 8606
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022062775502173877,
+      "loss": 0.0129,
+      "step": 8607
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022061065242087393,
+      "loss": 0.0087,
+      "step": 8608
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002205935486406778,
+      "loss": 0.0047,
+      "step": 8609
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002205764436814362,
+      "loss": 0.0096,
+      "step": 8610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022055933754343475,
+      "loss": 0.0061,
+      "step": 8611
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002205422302269592,
+      "loss": 0.0235,
+      "step": 8612
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022052512173229523,
+      "loss": 0.012,
+      "step": 8613
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022050801205972858,
+      "loss": 0.0147,
+      "step": 8614
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022049090120954503,
+      "loss": 0.0076,
+      "step": 8615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022047378918203038,
+      "loss": 0.0315,
+      "step": 8616
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022045667597747046,
+      "loss": 0.0158,
+      "step": 8617
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022043956159615102,
+      "loss": 0.0133,
+      "step": 8618
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022042244603835795,
+      "loss": 0.0256,
+      "step": 8619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022040532930437708,
+      "loss": 0.0133,
+      "step": 8620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022038821139449432,
+      "loss": 0.0184,
+      "step": 8621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022037109230899562,
+      "loss": 0.0138,
+      "step": 8622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022035397204816677,
+      "loss": 0.0201,
+      "step": 8623
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002203368506122938,
+      "loss": 0.0172,
+      "step": 8624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022031972800166269,
+      "loss": 0.0119,
+      "step": 8625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022030260421655932,
+      "loss": 0.0073,
+      "step": 8626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002202854792572698,
+      "loss": 0.017,
+      "step": 8627
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002202683531240801,
+      "loss": 0.0075,
+      "step": 8628
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022025122581727626,
+      "loss": 0.0161,
+      "step": 8629
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022023409733714432,
+      "loss": 0.0092,
+      "step": 8630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002202169676839704,
+      "loss": 0.0166,
+      "step": 8631
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022019983685804057,
+      "loss": 0.0048,
+      "step": 8632
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022018270485964095,
+      "loss": 0.0119,
+      "step": 8633
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002201655716890577,
+      "loss": 0.0115,
+      "step": 8634
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022014843734657693,
+      "loss": 0.0094,
+      "step": 8635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022013130183248482,
+      "loss": 0.0082,
+      "step": 8636
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022011416514706764,
+      "loss": 0.0188,
+      "step": 8637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002200970272906115,
+      "loss": 0.0078,
+      "step": 8638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022007988826340272,
+      "loss": 0.0023,
+      "step": 8639
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022006274806572756,
+      "loss": 0.0102,
+      "step": 8640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002200456066978722,
+      "loss": 0.0292,
+      "step": 8641
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022002846416012298,
+      "loss": 0.0513,
+      "step": 8642
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022001132045276626,
+      "loss": 0.0072,
+      "step": 8643
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002199941755760883,
+      "loss": 0.0053,
+      "step": 8644
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021997702953037548,
+      "loss": 0.0129,
+      "step": 8645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002199598823159142,
+      "loss": 0.0102,
+      "step": 8646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002199427339329908,
+      "loss": 0.0251,
+      "step": 8647
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021992558438189175,
+      "loss": 0.0172,
+      "step": 8648
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021990843366290346,
+      "loss": 0.0081,
+      "step": 8649
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021989128177631232,
+      "loss": 0.0185,
+      "step": 8650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021987412872240486,
+      "loss": 0.0222,
+      "step": 8651
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021985697450146758,
+      "loss": 0.0366,
+      "step": 8652
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002198398191137869,
+      "loss": 0.0109,
+      "step": 8653
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002198226625596495,
+      "loss": 0.0185,
+      "step": 8654
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021980550483934177,
+      "loss": 0.0176,
+      "step": 8655
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021978834595315033,
+      "loss": 0.0179,
+      "step": 8656
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002197711859013618,
+      "loss": 0.0156,
+      "step": 8657
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021975402468426276,
+      "loss": 0.0057,
+      "step": 8658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021973686230213985,
+      "loss": 0.014,
+      "step": 8659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021971969875527974,
+      "loss": 0.0267,
+      "step": 8660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021970253404396896,
+      "loss": 0.0142,
+      "step": 8661
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021968536816849437,
+      "loss": 0.0179,
+      "step": 8662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021966820112914257,
+      "loss": 0.0198,
+      "step": 8663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021965103292620027,
+      "loss": 0.0153,
+      "step": 8664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021963386355995426,
+      "loss": 0.0176,
+      "step": 8665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021961669303069133,
+      "loss": 0.0059,
+      "step": 8666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002195995213386982,
+      "loss": 0.01,
+      "step": 8667
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021958234848426164,
+      "loss": 0.0023,
+      "step": 8668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021956517446766852,
+      "loss": 0.0183,
+      "step": 8669
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021954799928920567,
+      "loss": 0.0063,
+      "step": 8670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021953082294916,
+      "loss": 0.0278,
+      "step": 8671
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002195136454478183,
+      "loss": 0.0174,
+      "step": 8672
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021949646678546755,
+      "loss": 0.0176,
+      "step": 8673
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021947928696239457,
+      "loss": 0.0198,
+      "step": 8674
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021946210597888636,
+      "loss": 0.0337,
+      "step": 8675
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021944492383522987,
+      "loss": 0.0124,
+      "step": 8676
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021942774053171205,
+      "loss": 0.0437,
+      "step": 8677
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002194105560686199,
+      "loss": 0.0228,
+      "step": 8678
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021939337044624046,
+      "loss": 0.0055,
+      "step": 8679
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021937618366486077,
+      "loss": 0.0052,
+      "step": 8680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021935899572476778,
+      "loss": 0.0108,
+      "step": 8681
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021934180662624875,
+      "loss": 0.0038,
+      "step": 8682
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002193246163695906,
+      "loss": 0.0148,
+      "step": 8683
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021930742495508048,
+      "loss": 0.0433,
+      "step": 8684
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021929023238300552,
+      "loss": 0.0132,
+      "step": 8685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021927303865365292,
+      "loss": 0.0046,
+      "step": 8686
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002192558437673098,
+      "loss": 0.0083,
+      "step": 8687
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021923864772426335,
+      "loss": 0.0091,
+      "step": 8688
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021922145052480084,
+      "loss": 0.0223,
+      "step": 8689
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021920425216920937,
+      "loss": 0.0042,
+      "step": 8690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191870526577763,
+      "loss": 0.0292,
+      "step": 8691
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021916985199078886,
+      "loss": 0.009,
+      "step": 8692
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191526501685343,
+      "loss": 0.0083,
+      "step": 8693
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191354471912999,
+      "loss": 0.0184,
+      "step": 8694
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191182430593731,
+      "loss": 0.0071,
+      "step": 8695
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191010377730412,
+      "loss": 0.0147,
+      "step": 8696
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002190838313325915,
+      "loss": 0.0035,
+      "step": 8697
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021906662373831135,
+      "loss": 0.0109,
+      "step": 8698
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021904941499048823,
+      "loss": 0.0177,
+      "step": 8699
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021903220508940962,
+      "loss": 0.0159,
+      "step": 8700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002190149940353628,
+      "loss": 0.027,
+      "step": 8701
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021899778182863532,
+      "loss": 0.0311,
+      "step": 8702
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021898056846951465,
+      "loss": 0.0186,
+      "step": 8703
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021896335395828827,
+      "loss": 0.0052,
+      "step": 8704
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021894613829524375,
+      "loss": 0.005,
+      "step": 8705
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002189289214806685,
+      "loss": 0.0175,
+      "step": 8706
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021891170351485015,
+      "loss": 0.016,
+      "step": 8707
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021889448439807633,
+      "loss": 0.0184,
+      "step": 8708
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021887726413063454,
+      "loss": 0.0217,
+      "step": 8709
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002188600427128124,
+      "loss": 0.0192,
+      "step": 8710
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002188428201448976,
+      "loss": 0.0398,
+      "step": 8711
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021882559642717777,
+      "loss": 0.0035,
+      "step": 8712
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021880837155994053,
+      "loss": 0.0108,
+      "step": 8713
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021879114554347362,
+      "loss": 0.0364,
+      "step": 8714
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021877391837806472,
+      "loss": 0.0135,
+      "step": 8715
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021875669006400154,
+      "loss": 0.0073,
+      "step": 8716
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021873946060157188,
+      "loss": 0.0123,
+      "step": 8717
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002187222299910635,
+      "loss": 0.0113,
+      "step": 8718
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002187049982327641,
+      "loss": 0.0171,
+      "step": 8719
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021868776532696152,
+      "loss": 0.0042,
+      "step": 8720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021867053127394364,
+      "loss": 0.0191,
+      "step": 8721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021865329607399824,
+      "loss": 0.0074,
+      "step": 8722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021863605972741328,
+      "loss": 0.0158,
+      "step": 8723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002186188222344765,
+      "loss": 0.0127,
+      "step": 8724
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021860158359547584,
+      "loss": 0.0248,
+      "step": 8725
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002185843438106993,
+      "loss": 0.0337,
+      "step": 8726
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021856710288043472,
+      "loss": 0.0278,
+      "step": 8727
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021854986080497007,
+      "loss": 0.0097,
+      "step": 8728
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021853261758459337,
+      "loss": 0.0085,
+      "step": 8729
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002185153732195926,
+      "loss": 0.025,
+      "step": 8730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021849812771025575,
+      "loss": 0.0082,
+      "step": 8731
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002184808810568709,
+      "loss": 0.0036,
+      "step": 8732
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021846363325972606,
+      "loss": 0.005,
+      "step": 8733
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021844638431910927,
+      "loss": 0.0123,
+      "step": 8734
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002184291342353087,
+      "loss": 0.0009,
+      "step": 8735
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021841188300861242,
+      "loss": 0.0102,
+      "step": 8736
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021839463063930858,
+      "loss": 0.0055,
+      "step": 8737
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021837737712768524,
+      "loss": 0.0336,
+      "step": 8738
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021836012247403066,
+      "loss": 0.1206,
+      "step": 8739
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000218342866678633,
+      "loss": 0.0128,
+      "step": 8740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002183256097417804,
+      "loss": 0.0221,
+      "step": 8741
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021830835166376125,
+      "loss": 0.0225,
+      "step": 8742
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021829109244486363,
+      "loss": 0.0359,
+      "step": 8743
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021827383208537586,
+      "loss": 0.0078,
+      "step": 8744
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021825657058558622,
+      "loss": 0.0117,
+      "step": 8745
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000218239307945783,
+      "loss": 0.0043,
+      "step": 8746
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021822204416625457,
+      "loss": 0.0081,
+      "step": 8747
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021820477924728917,
+      "loss": 0.0046,
+      "step": 8748
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021818751318917524,
+      "loss": 0.0121,
+      "step": 8749
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021817024599220112,
+      "loss": 0.0159,
+      "step": 8750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021815297765665517,
+      "loss": 0.0025,
+      "step": 8751
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002181357081828259,
+      "loss": 0.0265,
+      "step": 8752
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002181184375710017,
+      "loss": 0.0129,
+      "step": 8753
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021810116582147098,
+      "loss": 0.016,
+      "step": 8754
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021808389293452221,
+      "loss": 0.0085,
+      "step": 8755
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021806661891044397,
+      "loss": 0.0069,
+      "step": 8756
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002180493437495247,
+      "loss": 0.0196,
+      "step": 8757
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002180320674520529,
+      "loss": 0.0059,
+      "step": 8758
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021801479001831717,
+      "loss": 0.0179,
+      "step": 8759
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021799751144860606,
+      "loss": 0.058,
+      "step": 8760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021798023174320817,
+      "loss": 0.0037,
+      "step": 8761
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021796295090241205,
+      "loss": 0.006,
+      "step": 8762
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021794566892650638,
+      "loss": 0.0184,
+      "step": 8763
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021792838581577981,
+      "loss": 0.0059,
+      "step": 8764
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021791110157052095,
+      "loss": 0.0062,
+      "step": 8765
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021789381619101853,
+      "loss": 0.0295,
+      "step": 8766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021787652967756118,
+      "loss": 0.0166,
+      "step": 8767
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002178592420304377,
+      "loss": 0.0166,
+      "step": 8768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021784195324993673,
+      "loss": 0.0074,
+      "step": 8769
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021782466333634712,
+      "loss": 0.009,
+      "step": 8770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021780737228995758,
+      "loss": 0.0167,
+      "step": 8771
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021779008011105693,
+      "loss": 0.0097,
+      "step": 8772
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000217772786799934,
+      "loss": 0.0275,
+      "step": 8773
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002177554923568776,
+      "loss": 0.0111,
+      "step": 8774
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021773819678217658,
+      "loss": 0.0128,
+      "step": 8775
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021772090007611978,
+      "loss": 0.017,
+      "step": 8776
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021770360223899614,
+      "loss": 0.0189,
+      "step": 8777
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021768630327109452,
+      "loss": 0.0094,
+      "step": 8778
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002176690031727039,
+      "loss": 0.0024,
+      "step": 8779
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002176517019441132,
+      "loss": 0.0043,
+      "step": 8780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021763439958561131,
+      "loss": 0.0139,
+      "step": 8781
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021761709609748734,
+      "loss": 0.025,
+      "step": 8782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002175997914800302,
+      "loss": 0.0066,
+      "step": 8783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021758248573352897,
+      "loss": 0.0078,
+      "step": 8784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021756517885827262,
+      "loss": 0.0107,
+      "step": 8785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002175478708545503,
+      "loss": 0.0304,
+      "step": 8786
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021753056172265096,
+      "loss": 0.0096,
+      "step": 8787
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002175132514628638,
+      "loss": 0.0073,
+      "step": 8788
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021749594007547794,
+      "loss": 0.0077,
+      "step": 8789
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021747862756078243,
+      "loss": 0.0322,
+      "step": 8790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021746131391906645,
+      "loss": 0.0092,
+      "step": 8791
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002174439991506192,
+      "loss": 0.0035,
+      "step": 8792
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021742668325572985,
+      "loss": 0.0213,
+      "step": 8793
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021740936623468766,
+      "loss": 0.0116,
+      "step": 8794
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021739204808778173,
+      "loss": 0.0304,
+      "step": 8795
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021737472881530143,
+      "loss": 0.0134,
+      "step": 8796
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021735740841753598,
+      "loss": 0.0115,
+      "step": 8797
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021734008689477466,
+      "loss": 0.0411,
+      "step": 8798
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021732276424730681,
+      "loss": 0.0095,
+      "step": 8799
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021730544047542167,
+      "loss": 0.0059,
+      "step": 8800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021728811557940862,
+      "loss": 0.0144,
+      "step": 8801
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021727078955955704,
+      "loss": 0.0161,
+      "step": 8802
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002172534624161563,
+      "loss": 0.0335,
+      "step": 8803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021723613414949578,
+      "loss": 0.0047,
+      "step": 8804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002172188047598649,
+      "loss": 0.0263,
+      "step": 8805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021720147424755308,
+      "loss": 0.0108,
+      "step": 8806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021718414261284982,
+      "loss": 0.0073,
+      "step": 8807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021716680985604459,
+      "loss": 0.0237,
+      "step": 8808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021714947597742677,
+      "loss": 0.0115,
+      "step": 8809
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021713214097728598,
+      "loss": 0.029,
+      "step": 8810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021711480485591173,
+      "loss": 0.0033,
+      "step": 8811
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021709746761359354,
+      "loss": 0.036,
+      "step": 8812
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021708012925062095,
+      "loss": 0.0046,
+      "step": 8813
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021706278976728365,
+      "loss": 0.0072,
+      "step": 8814
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021704544916387113,
+      "loss": 0.0144,
+      "step": 8815
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021702810744067308,
+      "loss": 0.0044,
+      "step": 8816
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021701076459797907,
+      "loss": 0.0068,
+      "step": 8817
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021699342063607883,
+      "loss": 0.0143,
+      "step": 8818
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021697607555526196,
+      "loss": 0.0335,
+      "step": 8819
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021695872935581826,
+      "loss": 0.0271,
+      "step": 8820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002169413820380374,
+      "loss": 0.0403,
+      "step": 8821
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021692403360220905,
+      "loss": 0.0191,
+      "step": 8822
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021690668404862295,
+      "loss": 0.0291,
+      "step": 8823
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021688933337756902,
+      "loss": 0.0031,
+      "step": 8824
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021687198158933696,
+      "loss": 0.0071,
+      "step": 8825
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021685462868421655,
+      "loss": 0.0167,
+      "step": 8826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021683727466249769,
+      "loss": 0.0013,
+      "step": 8827
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021681991952447012,
+      "loss": 0.0031,
+      "step": 8828
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021680256327042375,
+      "loss": 0.0048,
+      "step": 8829
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021678520590064855,
+      "loss": 0.0121,
+      "step": 8830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002167678474154343,
+      "loss": 0.0167,
+      "step": 8831
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021675048781507095,
+      "loss": 0.0327,
+      "step": 8832
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021673312709984848,
+      "loss": 0.0071,
+      "step": 8833
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021671576527005682,
+      "loss": 0.0056,
+      "step": 8834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021669840232598595,
+      "loss": 0.0224,
+      "step": 8835
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021668103826792584,
+      "loss": 0.0111,
+      "step": 8836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021666367309616653,
+      "loss": 0.0157,
+      "step": 8837
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021664630681099802,
+      "loss": 0.0294,
+      "step": 8838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021662893941271045,
+      "loss": 0.0081,
+      "step": 8839
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021661157090159378,
+      "loss": 0.0171,
+      "step": 8840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021659420127793813,
+      "loss": 0.0077,
+      "step": 8841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002165768305420336,
+      "loss": 0.0158,
+      "step": 8842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002165594586941703,
+      "loss": 0.0203,
+      "step": 8843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021654208573463845,
+      "loss": 0.012,
+      "step": 8844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021652471166372817,
+      "loss": 0.0219,
+      "step": 8845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002165073364817296,
+      "loss": 0.006,
+      "step": 8846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021648996018893302,
+      "loss": 0.0027,
+      "step": 8847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021647258278562856,
+      "loss": 0.0114,
+      "step": 8848
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021645520427210647,
+      "loss": 0.0083,
+      "step": 8849
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021643782464865707,
+      "loss": 0.0104,
+      "step": 8850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021642044391557055,
+      "loss": 0.007,
+      "step": 8851
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021640306207313723,
+      "loss": 0.015,
+      "step": 8852
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002163856791216475,
+      "loss": 0.0048,
+      "step": 8853
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021636829506139154,
+      "loss": 0.0156,
+      "step": 8854
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021635090989265978,
+      "loss": 0.0129,
+      "step": 8855
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021633352361574262,
+      "loss": 0.0084,
+      "step": 8856
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021631613623093038,
+      "loss": 0.035,
+      "step": 8857
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002162987477385135,
+      "loss": 0.0161,
+      "step": 8858
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021628135813878238,
+      "loss": 0.0358,
+      "step": 8859
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021626396743202746,
+      "loss": 0.008,
+      "step": 8860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021624657561853918,
+      "loss": 0.0134,
+      "step": 8861
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002162291826986081,
+      "loss": 0.0067,
+      "step": 8862
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021621178867252462,
+      "loss": 0.0159,
+      "step": 8863
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021619439354057923,
+      "loss": 0.0196,
+      "step": 8864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021617699730306256,
+      "loss": 0.0285,
+      "step": 8865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021615959996026512,
+      "loss": 0.013,
+      "step": 8866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021614220151247747,
+      "loss": 0.012,
+      "step": 8867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021612480195999024,
+      "loss": 0.0075,
+      "step": 8868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021610740130309397,
+      "loss": 0.0106,
+      "step": 8869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021608999954207927,
+      "loss": 0.0254,
+      "step": 8870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002160725966772369,
+      "loss": 0.0629,
+      "step": 8871
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002160551927088574,
+      "loss": 0.0462,
+      "step": 8872
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002160377876372315,
+      "loss": 0.0063,
+      "step": 8873
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021602038146264991,
+      "loss": 0.0148,
+      "step": 8874
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021600297418540327,
+      "loss": 0.0032,
+      "step": 8875
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021598556580578243,
+      "loss": 0.0025,
+      "step": 8876
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021596815632407803,
+      "loss": 0.0164,
+      "step": 8877
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021595074574058093,
+      "loss": 0.0418,
+      "step": 8878
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002159333340555819,
+      "loss": 0.0067,
+      "step": 8879
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002159159212693717,
+      "loss": 0.0201,
+      "step": 8880
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021589850738224127,
+      "loss": 0.0053,
+      "step": 8881
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002158810923944813,
+      "loss": 0.014,
+      "step": 8882
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002158636763063827,
+      "loss": 0.0136,
+      "step": 8883
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021584625911823642,
+      "loss": 0.018,
+      "step": 8884
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021582884083033333,
+      "loss": 0.0143,
+      "step": 8885
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021581142144296432,
+      "loss": 0.0129,
+      "step": 8886
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021579400095642036,
+      "loss": 0.025,
+      "step": 8887
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021577657937099236,
+      "loss": 0.0072,
+      "step": 8888
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021575915668697136,
+      "loss": 0.0084,
+      "step": 8889
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002157417329046483,
+      "loss": 0.0121,
+      "step": 8890
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002157243080243142,
+      "loss": 0.0087,
+      "step": 8891
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021570688204626008,
+      "loss": 0.0152,
+      "step": 8892
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000215689454970777,
+      "loss": 0.007,
+      "step": 8893
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021567202679815604,
+      "loss": 0.015,
+      "step": 8894
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021565459752868828,
+      "loss": 0.0253,
+      "step": 8895
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021563716716266478,
+      "loss": 0.0063,
+      "step": 8896
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002156197357003767,
+      "loss": 0.0162,
+      "step": 8897
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021560230314211518,
+      "loss": 0.0165,
+      "step": 8898
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021558486948817138,
+      "loss": 0.005,
+      "step": 8899
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021556743473883643,
+      "loss": 0.0017,
+      "step": 8900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021554999889440154,
+      "loss": 0.0069,
+      "step": 8901
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021553256195515794,
+      "loss": 0.0147,
+      "step": 8902
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021551512392139688,
+      "loss": 0.0155,
+      "step": 8903
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021549768479340957,
+      "loss": 0.0094,
+      "step": 8904
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002154802445714873,
+      "loss": 0.0057,
+      "step": 8905
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021546280325592128,
+      "loss": 0.0052,
+      "step": 8906
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021544536084700295,
+      "loss": 0.0161,
+      "step": 8907
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021542791734502356,
+      "loss": 0.0207,
+      "step": 8908
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002154104727502744,
+      "loss": 0.0203,
+      "step": 8909
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021539302706304687,
+      "loss": 0.0066,
+      "step": 8910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002153755802836324,
+      "loss": 0.0149,
+      "step": 8911
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021535813241232226,
+      "loss": 0.017,
+      "step": 8912
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021534068344940805,
+      "loss": 0.0174,
+      "step": 8913
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000215323233395181,
+      "loss": 0.032,
+      "step": 8914
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021530578224993263,
+      "loss": 0.0041,
+      "step": 8915
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021528833001395448,
+      "loss": 0.0128,
+      "step": 8916
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021527087668753793,
+      "loss": 0.0057,
+      "step": 8917
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021525342227097458,
+      "loss": 0.0089,
+      "step": 8918
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021523596676455587,
+      "loss": 0.0054,
+      "step": 8919
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002152185101685734,
+      "loss": 0.0116,
+      "step": 8920
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002152010524833187,
+      "loss": 0.0073,
+      "step": 8921
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002151835937090833,
+      "loss": 0.0102,
+      "step": 8922
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002151661338461589,
+      "loss": 0.0245,
+      "step": 8923
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021514867289483704,
+      "loss": 0.0303,
+      "step": 8924
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021513121085540933,
+      "loss": 0.0245,
+      "step": 8925
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002151137477281675,
+      "loss": 0.0055,
+      "step": 8926
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021509628351340315,
+      "loss": 0.0245,
+      "step": 8927
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000215078818211408,
+      "loss": 0.002,
+      "step": 8928
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002150613518224737,
+      "loss": 0.0153,
+      "step": 8929
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021504388434689204,
+      "loss": 0.0079,
+      "step": 8930
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021502641578495475,
+      "loss": 0.0195,
+      "step": 8931
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002150089461369536,
+      "loss": 0.0213,
+      "step": 8932
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021499147540318027,
+      "loss": 0.0452,
+      "step": 8933
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021497400358392663,
+      "loss": 0.0029,
+      "step": 8934
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021495653067948452,
+      "loss": 0.0086,
+      "step": 8935
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002149390566901457,
+      "loss": 0.0124,
+      "step": 8936
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021492158161620206,
+      "loss": 0.0133,
+      "step": 8937
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021490410545794546,
+      "loss": 0.0171,
+      "step": 8938
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002148866282156678,
+      "loss": 0.0182,
+      "step": 8939
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021486914988966095,
+      "loss": 0.0031,
+      "step": 8940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021485167048021685,
+      "loss": 0.0066,
+      "step": 8941
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021483418998762742,
+      "loss": 0.0058,
+      "step": 8942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021481670841218465,
+      "loss": 0.0072,
+      "step": 8943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021479922575418051,
+      "loss": 0.0045,
+      "step": 8944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021478174201390699,
+      "loss": 0.0122,
+      "step": 8945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021476425719165612,
+      "loss": 0.0369,
+      "step": 8946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002147467712877198,
+      "loss": 0.0112,
+      "step": 8947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021472928430239023,
+      "loss": 0.0033,
+      "step": 8948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021471179623595943,
+      "loss": 0.0112,
+      "step": 8949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021469430708871953,
+      "loss": 0.0031,
+      "step": 8950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021467681686096254,
+      "loss": 0.0201,
+      "step": 8951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021465932555298062,
+      "loss": 0.0099,
+      "step": 8952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002146418331650659,
+      "loss": 0.0218,
+      "step": 8953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021462433969751058,
+      "loss": 0.0078,
+      "step": 8954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002146068451506068,
+      "loss": 0.0313,
+      "step": 8955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021458934952464672,
+      "loss": 0.0018,
+      "step": 8956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021457185281992258,
+      "loss": 0.0038,
+      "step": 8957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021455435503672665,
+      "loss": 0.0079,
+      "step": 8958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021453685617535108,
+      "loss": 0.048,
+      "step": 8959
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021451935623608822,
+      "loss": 0.0088,
+      "step": 8960
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002145018552192303,
+      "loss": 0.0098,
+      "step": 8961
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002144843531250697,
+      "loss": 0.0097,
+      "step": 8962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021446684995389865,
+      "loss": 0.0022,
+      "step": 8963
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021444934570600952,
+      "loss": 0.0127,
+      "step": 8964
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021443184038169466,
+      "loss": 0.0327,
+      "step": 8965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002144143339812464,
+      "loss": 0.0028,
+      "step": 8966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002143968265049572,
+      "loss": 0.0036,
+      "step": 8967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021437931795311946,
+      "loss": 0.0079,
+      "step": 8968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021436180832602554,
+      "loss": 0.0707,
+      "step": 8969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000214344297623968,
+      "loss": 0.0352,
+      "step": 8970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021432678584723915,
+      "loss": 0.0053,
+      "step": 8971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002143092729961316,
+      "loss": 0.0139,
+      "step": 8972
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002142917590709378,
+      "loss": 0.0395,
+      "step": 8973
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021427424407195026,
+      "loss": 0.0114,
+      "step": 8974
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002142567279994615,
+      "loss": 0.0359,
+      "step": 8975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002142392108537641,
+      "loss": 0.0119,
+      "step": 8976
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002142216926351506,
+      "loss": 0.0207,
+      "step": 8977
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021420417334391364,
+      "loss": 0.0225,
+      "step": 8978
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021418665298034576,
+      "loss": 0.0279,
+      "step": 8979
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021416913154473965,
+      "loss": 0.0129,
+      "step": 8980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002141516090373879,
+      "loss": 0.0346,
+      "step": 8981
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002141340854585832,
+      "loss": 0.0274,
+      "step": 8982
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021411656080861817,
+      "loss": 0.0016,
+      "step": 8983
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021409903508778558,
+      "loss": 0.0148,
+      "step": 8984
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021408150829637808,
+      "loss": 0.0086,
+      "step": 8985
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021406398043468845,
+      "loss": 0.0103,
+      "step": 8986
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021404645150300945,
+      "loss": 0.0264,
+      "step": 8987
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021402892150163377,
+      "loss": 0.0082,
+      "step": 8988
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002140113904308542,
+      "loss": 0.0035,
+      "step": 8989
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002139938582909636,
+      "loss": 0.0133,
+      "step": 8990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021397632508225482,
+      "loss": 0.0116,
+      "step": 8991
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021395879080502066,
+      "loss": 0.0284,
+      "step": 8992
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021394125545955392,
+      "loss": 0.0218,
+      "step": 8993
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021392371904614753,
+      "loss": 0.025,
+      "step": 8994
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021390618156509435,
+      "loss": 0.0064,
+      "step": 8995
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002138886430166873,
+      "loss": 0.0247,
+      "step": 8996
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021387110340121934,
+      "loss": 0.0032,
+      "step": 8997
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021385356271898338,
+      "loss": 0.0324,
+      "step": 8998
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021383602097027238,
+      "loss": 0.0152,
+      "step": 8999
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021381847815537928,
+      "loss": 0.0149,
+      "step": 9000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021380093427459724,
+      "loss": 0.0116,
+      "step": 9001
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002137833893282191,
+      "loss": 0.0062,
+      "step": 9002
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021376584331653795,
+      "loss": 0.004,
+      "step": 9003
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021374829623984685,
+      "loss": 0.0055,
+      "step": 9004
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021373074809843888,
+      "loss": 0.0214,
+      "step": 9005
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021371319889260717,
+      "loss": 0.019,
+      "step": 9006
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002136956486226447,
+      "loss": 0.0015,
+      "step": 9007
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002136780972888447,
+      "loss": 0.003,
+      "step": 9008
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021366054489150029,
+      "loss": 0.0068,
+      "step": 9009
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021364299143090455,
+      "loss": 0.011,
+      "step": 9010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021362543690735077,
+      "loss": 0.0121,
+      "step": 9011
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021360788132113206,
+      "loss": 0.0101,
+      "step": 9012
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002135903246725417,
+      "loss": 0.0078,
+      "step": 9013
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021357276696187284,
+      "loss": 0.0043,
+      "step": 9014
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002135552081894188,
+      "loss": 0.0107,
+      "step": 9015
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021353764835547282,
+      "loss": 0.0102,
+      "step": 9016
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021352008746032815,
+      "loss": 0.0086,
+      "step": 9017
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021350252550427816,
+      "loss": 0.0303,
+      "step": 9018
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002134849624876161,
+      "loss": 0.0117,
+      "step": 9019
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021346739841063527,
+      "loss": 0.0099,
+      "step": 9020
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021344983327362916,
+      "loss": 0.0118,
+      "step": 9021
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021343226707689104,
+      "loss": 0.0212,
+      "step": 9022
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021341469982071433,
+      "loss": 0.0225,
+      "step": 9023
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021339713150539245,
+      "loss": 0.0152,
+      "step": 9024
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021337956213121878,
+      "loss": 0.0173,
+      "step": 9025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021336199169848678,
+      "loss": 0.0114,
+      "step": 9026
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002133444202074899,
+      "loss": 0.0035,
+      "step": 9027
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021332684765852165,
+      "loss": 0.0203,
+      "step": 9028
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021330927405187553,
+      "loss": 0.0111,
+      "step": 9029
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021329169938784498,
+      "loss": 0.0109,
+      "step": 9030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002132741236667236,
+      "loss": 0.0238,
+      "step": 9031
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021325654688880488,
+      "loss": 0.0136,
+      "step": 9032
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002132389690543825,
+      "loss": 0.0401,
+      "step": 9033
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021322139016374992,
+      "loss": 0.008,
+      "step": 9034
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002132038102172008,
+      "loss": 0.0082,
+      "step": 9035
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021318622921502871,
+      "loss": 0.0098,
+      "step": 9036
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021316864715752737,
+      "loss": 0.0024,
+      "step": 9037
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021315106404499036,
+      "loss": 0.0095,
+      "step": 9038
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002131334798777114,
+      "loss": 0.0236,
+      "step": 9039
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021311589465598408,
+      "loss": 0.0237,
+      "step": 9040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021309830838010224,
+      "loss": 0.0436,
+      "step": 9041
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002130807210503595,
+      "loss": 0.0148,
+      "step": 9042
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002130631326670497,
+      "loss": 0.0032,
+      "step": 9043
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002130455432304665,
+      "loss": 0.0098,
+      "step": 9044
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021302795274090374,
+      "loss": 0.0126,
+      "step": 9045
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021301036119865518,
+      "loss": 0.0092,
+      "step": 9046
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021299276860401467,
+      "loss": 0.0079,
+      "step": 9047
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021297517495727603,
+      "loss": 0.02,
+      "step": 9048
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021295758025873307,
+      "loss": 0.0076,
+      "step": 9049
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021293998450867967,
+      "loss": 0.0159,
+      "step": 9050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021292238770740973,
+      "loss": 0.0145,
+      "step": 9051
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021290478985521718,
+      "loss": 0.0074,
+      "step": 9052
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021288719095239583,
+      "loss": 0.0164,
+      "step": 9053
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021286959099923972,
+      "loss": 0.0227,
+      "step": 9054
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021285198999604273,
+      "loss": 0.006,
+      "step": 9055
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021283438794309894,
+      "loss": 0.0095,
+      "step": 9056
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021281678484070222,
+      "loss": 0.0155,
+      "step": 9057
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021279918068914662,
+      "loss": 0.0159,
+      "step": 9058
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021278157548872614,
+      "loss": 0.0305,
+      "step": 9059
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021276396923973487,
+      "loss": 0.0087,
+      "step": 9060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021274636194246681,
+      "loss": 0.0154,
+      "step": 9061
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002127287535972161,
+      "loss": 0.0183,
+      "step": 9062
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021271114420427675,
+      "loss": 0.0125,
+      "step": 9063
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021269353376394295,
+      "loss": 0.012,
+      "step": 9064
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021267592227650878,
+      "loss": 0.0012,
+      "step": 9065
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021265830974226844,
+      "loss": 0.0035,
+      "step": 9066
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021264069616151602,
+      "loss": 0.0417,
+      "step": 9067
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021262308153454572,
+      "loss": 0.007,
+      "step": 9068
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021260546586165173,
+      "loss": 0.0118,
+      "step": 9069
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021258784914312835,
+      "loss": 0.0115,
+      "step": 9070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002125702313792697,
+      "loss": 0.0114,
+      "step": 9071
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021255261257037007,
+      "loss": 0.0044,
+      "step": 9072
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021253499271672375,
+      "loss": 0.0061,
+      "step": 9073
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021251737181862497,
+      "loss": 0.0171,
+      "step": 9074
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021249974987636814,
+      "loss": 0.0057,
+      "step": 9075
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021248212689024747,
+      "loss": 0.0201,
+      "step": 9076
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021246450286055733,
+      "loss": 0.0071,
+      "step": 9077
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021244687778759208,
+      "loss": 0.012,
+      "step": 9078
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021242925167164612,
+      "loss": 0.0208,
+      "step": 9079
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021241162451301384,
+      "loss": 0.0097,
+      "step": 9080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021239399631198957,
+      "loss": 0.0013,
+      "step": 9081
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002123763670688678,
+      "loss": 0.0116,
+      "step": 9082
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021235873678394294,
+      "loss": 0.0018,
+      "step": 9083
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002123411054575095,
+      "loss": 0.0112,
+      "step": 9084
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021232347308986187,
+      "loss": 0.019,
+      "step": 9085
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021230583968129464,
+      "loss": 0.0117,
+      "step": 9086
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002122882052321023,
+      "loss": 0.016,
+      "step": 9087
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002122705697425793,
+      "loss": 0.0406,
+      "step": 9088
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021225293321302025,
+      "loss": 0.0194,
+      "step": 9089
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021223529564371973,
+      "loss": 0.0097,
+      "step": 9090
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021221765703497228,
+      "loss": 0.0338,
+      "step": 9091
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002122000173870725,
+      "loss": 0.0161,
+      "step": 9092
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000212182376700315,
+      "loss": 0.0248,
+      "step": 9093
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021216473497499448,
+      "loss": 0.0081,
+      "step": 9094
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021214709221140554,
+      "loss": 0.0168,
+      "step": 9095
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021212944840984284,
+      "loss": 0.0066,
+      "step": 9096
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021211180357060108,
+      "loss": 0.0164,
+      "step": 9097
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021209415769397495,
+      "loss": 0.0053,
+      "step": 9098
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021207651078025916,
+      "loss": 0.0092,
+      "step": 9099
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021205886282974848,
+      "loss": 0.0063,
+      "step": 9100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021204121384273762,
+      "loss": 0.0071,
+      "step": 9101
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002120235638195214,
+      "loss": 0.0145,
+      "step": 9102
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021200591276039458,
+      "loss": 0.0136,
+      "step": 9103
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021198826066565196,
+      "loss": 0.0076,
+      "step": 9104
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002119706075355884,
+      "loss": 0.037,
+      "step": 9105
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002119529533704987,
+      "loss": 0.0168,
+      "step": 9106
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021193529817067777,
+      "loss": 0.0275,
+      "step": 9107
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002119176419364204,
+      "loss": 0.014,
+      "step": 9108
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021189998466802158,
+      "loss": 0.0066,
+      "step": 9109
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021188232636577614,
+      "loss": 0.0044,
+      "step": 9110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021186466702997905,
+      "loss": 0.006,
+      "step": 9111
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021184700666092526,
+      "loss": 0.0098,
+      "step": 9112
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002118293452589097,
+      "loss": 0.0099,
+      "step": 9113
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002118116828242273,
+      "loss": 0.007,
+      "step": 9114
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021179401935717316,
+      "loss": 0.0237,
+      "step": 9115
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021177635485804226,
+      "loss": 0.0173,
+      "step": 9116
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021175868932712968,
+      "loss": 0.0095,
+      "step": 9117
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021174102276473036,
+      "loss": 0.0081,
+      "step": 9118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021172335517113936,
+      "loss": 0.0096,
+      "step": 9119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021170568654665188,
+      "loss": 0.017,
+      "step": 9120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000211688016891563,
+      "loss": 0.0207,
+      "step": 9121
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021167034620616772,
+      "loss": 0.012,
+      "step": 9122
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021165267449076128,
+      "loss": 0.0028,
+      "step": 9123
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002116350017456388,
+      "loss": 0.0024,
+      "step": 9124
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002116173279710954,
+      "loss": 0.025,
+      "step": 9125
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021159965316742632,
+      "loss": 0.008,
+      "step": 9126
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002115819773349268,
+      "loss": 0.0104,
+      "step": 9127
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000211564300473892,
+      "loss": 0.0416,
+      "step": 9128
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021154662258461717,
+      "loss": 0.0072,
+      "step": 9129
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021152894366739753,
+      "loss": 0.006,
+      "step": 9130
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021151126372252846,
+      "loss": 0.0106,
+      "step": 9131
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002114935827503051,
+      "loss": 0.0091,
+      "step": 9132
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021147590075102287,
+      "loss": 0.0225,
+      "step": 9133
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021145821772497705,
+      "loss": 0.0034,
+      "step": 9134
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021144053367246296,
+      "loss": 0.0215,
+      "step": 9135
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021142284859377601,
+      "loss": 0.0106,
+      "step": 9136
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021140516248921153,
+      "loss": 0.0262,
+      "step": 9137
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021138747535906494,
+      "loss": 0.0144,
+      "step": 9138
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021136978720363161,
+      "loss": 0.0186,
+      "step": 9139
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021135209802320706,
+      "loss": 0.0064,
+      "step": 9140
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021133440781808657,
+      "loss": 0.0106,
+      "step": 9141
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021131671658856574,
+      "loss": 0.0055,
+      "step": 9142
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021129902433493999,
+      "loss": 0.0265,
+      "step": 9143
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002112813310575048,
+      "loss": 0.0075,
+      "step": 9144
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021126363675655574,
+      "loss": 0.0236,
+      "step": 9145
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002112459414323883,
+      "loss": 0.0096,
+      "step": 9146
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021122824508529802,
+      "loss": 0.0055,
+      "step": 9147
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021121054771558047,
+      "loss": 0.0107,
+      "step": 9148
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021119284932353123,
+      "loss": 0.0255,
+      "step": 9149
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021117514990944588,
+      "loss": 0.0063,
+      "step": 9150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021115744947362006,
+      "loss": 0.0032,
+      "step": 9151
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021113974801634944,
+      "loss": 0.0083,
+      "step": 9152
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021112204553792955,
+      "loss": 0.0254,
+      "step": 9153
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021110434203865618,
+      "loss": 0.0044,
+      "step": 9154
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002110866375188249,
+      "loss": 0.035,
+      "step": 9155
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021106893197873149,
+      "loss": 0.0078,
+      "step": 9156
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021105122541867163,
+      "loss": 0.0028,
+      "step": 9157
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002110335178389411,
+      "loss": 0.0194,
+      "step": 9158
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021101580923983557,
+      "loss": 0.0054,
+      "step": 9159
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021099809962165088,
+      "loss": 0.0106,
+      "step": 9160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021098038898468275,
+      "loss": 0.0063,
+      "step": 9161
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021096267732922705,
+      "loss": 0.0076,
+      "step": 9162
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021094496465557958,
+      "loss": 0.0372,
+      "step": 9163
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002109272509640361,
+      "loss": 0.0254,
+      "step": 9164
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021090953625489255,
+      "loss": 0.0028,
+      "step": 9165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021089182052844477,
+      "loss": 0.0054,
+      "step": 9166
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002108741037849886,
+      "loss": 0.008,
+      "step": 9167
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021085638602482006,
+      "loss": 0.0097,
+      "step": 9168
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000210838667248235,
+      "loss": 0.0275,
+      "step": 9169
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002108209474555293,
+      "loss": 0.0137,
+      "step": 9170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021080322664699901,
+      "loss": 0.0045,
+      "step": 9171
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021078550482294008,
+      "loss": 0.0265,
+      "step": 9172
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021076778198364843,
+      "loss": 0.0174,
+      "step": 9173
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002107500581294201,
+      "loss": 0.0099,
+      "step": 9174
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021073233326055117,
+      "loss": 0.0112,
+      "step": 9175
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002107146073773376,
+      "loss": 0.0156,
+      "step": 9176
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002106968804800755,
+      "loss": 0.0065,
+      "step": 9177
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021067915256906095,
+      "loss": 0.0146,
+      "step": 9178
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021066142364458998,
+      "loss": 0.0099,
+      "step": 9179
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002106436937069587,
+      "loss": 0.0183,
+      "step": 9180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021062596275646332,
+      "loss": 0.0084,
+      "step": 9181
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021060823079339992,
+      "loss": 0.0042,
+      "step": 9182
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002105904978180646,
+      "loss": 0.0081,
+      "step": 9183
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021057276383075363,
+      "loss": 0.0025,
+      "step": 9184
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002105550288317632,
+      "loss": 0.0051,
+      "step": 9185
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021053729282138943,
+      "loss": 0.0618,
+      "step": 9186
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002105195557999286,
+      "loss": 0.0158,
+      "step": 9187
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021050181776767696,
+      "loss": 0.0116,
+      "step": 9188
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021048407872493075,
+      "loss": 0.0077,
+      "step": 9189
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002104663386719863,
+      "loss": 0.0075,
+      "step": 9190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021044859760913982,
+      "loss": 0.0155,
+      "step": 9191
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021043085553668764,
+      "loss": 0.0338,
+      "step": 9192
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002104131124549261,
+      "loss": 0.0086,
+      "step": 9193
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002103953683641516,
+      "loss": 0.0255,
+      "step": 9194
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021037762326466043,
+      "loss": 0.0167,
+      "step": 9195
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021035987715674895,
+      "loss": 0.0264,
+      "step": 9196
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021034213004071355,
+      "loss": 0.0124,
+      "step": 9197
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002103243819168507,
+      "loss": 0.0162,
+      "step": 9198
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021030663278545685,
+      "loss": 0.0124,
+      "step": 9199
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021028888264682834,
+      "loss": 0.0098,
+      "step": 9200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021027113150126172,
+      "loss": 0.0055,
+      "step": 9201
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021025337934905343,
+      "loss": 0.0069,
+      "step": 9202
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021023562619049997,
+      "loss": 0.0204,
+      "step": 9203
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021021787202589782,
+      "loss": 0.023,
+      "step": 9204
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002102001168555436,
+      "loss": 0.0175,
+      "step": 9205
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021018236067973375,
+      "loss": 0.0065,
+      "step": 9206
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021016460349876487,
+      "loss": 0.0102,
+      "step": 9207
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002101468453129335,
+      "loss": 0.0035,
+      "step": 9208
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021012908612253638,
+      "loss": 0.0054,
+      "step": 9209
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021011132592786995,
+      "loss": 0.0168,
+      "step": 9210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021009356472923092,
+      "loss": 0.0085,
+      "step": 9211
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021007580252691596,
+      "loss": 0.0233,
+      "step": 9212
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021005803932122163,
+      "loss": 0.0105,
+      "step": 9213
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021004027511244475,
+      "loss": 0.0065,
+      "step": 9214
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002100225099008819,
+      "loss": 0.0174,
+      "step": 9215
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021000474368682985,
+      "loss": 0.0126,
+      "step": 9216
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002099869764705853,
+      "loss": 0.0092,
+      "step": 9217
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020996920825244497,
+      "loss": 0.0087,
+      "step": 9218
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020995143903270575,
+      "loss": 0.0111,
+      "step": 9219
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002099336688116643,
+      "loss": 0.0343,
+      "step": 9220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002099158975896174,
+      "loss": 0.0128,
+      "step": 9221
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002098981253668619,
+      "loss": 0.0089,
+      "step": 9222
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020988035214369475,
+      "loss": 0.004,
+      "step": 9223
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020986257792041262,
+      "loss": 0.0234,
+      "step": 9224
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020984480269731242,
+      "loss": 0.0126,
+      "step": 9225
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020982702647469106,
+      "loss": 0.0106,
+      "step": 9226
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020980924925284548,
+      "loss": 0.0269,
+      "step": 9227
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020979147103207248,
+      "loss": 0.0107,
+      "step": 9228
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020977369181266907,
+      "loss": 0.0084,
+      "step": 9229
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020975591159493216,
+      "loss": 0.0063,
+      "step": 9230
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020973813037915875,
+      "loss": 0.0098,
+      "step": 9231
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002097203481656458,
+      "loss": 0.0325,
+      "step": 9232
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020970256495469032,
+      "loss": 0.0044,
+      "step": 9233
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020968478074658928,
+      "loss": 0.0024,
+      "step": 9234
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020966699554163975,
+      "loss": 0.0168,
+      "step": 9235
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020964920934013877,
+      "loss": 0.0065,
+      "step": 9236
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020963142214238339,
+      "loss": 0.0044,
+      "step": 9237
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002096136339486707,
+      "loss": 0.0169,
+      "step": 9238
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020959584475929775,
+      "loss": 0.0103,
+      "step": 9239
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020957805457456173,
+      "loss": 0.007,
+      "step": 9240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020956026339475974,
+      "loss": 0.0472,
+      "step": 9241
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020954247122018896,
+      "loss": 0.0277,
+      "step": 9242
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020952467805114647,
+      "loss": 0.0048,
+      "step": 9243
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020950688388792952,
+      "loss": 0.0111,
+      "step": 9244
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020948908873083527,
+      "loss": 0.0193,
+      "step": 9245
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020947129258016097,
+      "loss": 0.0113,
+      "step": 9246
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020945349543620382,
+      "loss": 0.0274,
+      "step": 9247
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020943569729926103,
+      "loss": 0.0209,
+      "step": 9248
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020941789816962992,
+      "loss": 0.011,
+      "step": 9249
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020940009804760773,
+      "loss": 0.0041,
+      "step": 9250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002093822969334918,
+      "loss": 0.0214,
+      "step": 9251
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020936449482757943,
+      "loss": 0.0506,
+      "step": 9252
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020934669173016792,
+      "loss": 0.0219,
+      "step": 9253
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020932888764155466,
+      "loss": 0.003,
+      "step": 9254
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000209311082562037,
+      "loss": 0.0102,
+      "step": 9255
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020929327649191223,
+      "loss": 0.0061,
+      "step": 9256
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020927546943147788,
+      "loss": 0.0229,
+      "step": 9257
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020925766138103126,
+      "loss": 0.0056,
+      "step": 9258
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020923985234086987,
+      "loss": 0.0073,
+      "step": 9259
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002092220423112911,
+      "loss": 0.0061,
+      "step": 9260
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020920423129259242,
+      "loss": 0.0026,
+      "step": 9261
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020918641928507136,
+      "loss": 0.01,
+      "step": 9262
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020916860628902536,
+      "loss": 0.0077,
+      "step": 9263
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020915079230475193,
+      "loss": 0.0041,
+      "step": 9264
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020913297733254866,
+      "loss": 0.0034,
+      "step": 9265
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020911516137271302,
+      "loss": 0.0249,
+      "step": 9266
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002090973444255426,
+      "loss": 0.0148,
+      "step": 9267
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020907952649133497,
+      "loss": 0.0169,
+      "step": 9268
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002090617075703877,
+      "loss": 0.0236,
+      "step": 9269
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020904388766299846,
+      "loss": 0.0217,
+      "step": 9270
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020902606676946484,
+      "loss": 0.0181,
+      "step": 9271
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020900824489008448,
+      "loss": 0.0289,
+      "step": 9272
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020899042202515505,
+      "loss": 0.0054,
+      "step": 9273
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002089725981749742,
+      "loss": 0.0193,
+      "step": 9274
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020895477333983967,
+      "loss": 0.0115,
+      "step": 9275
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020893694752004914,
+      "loss": 0.0223,
+      "step": 9276
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002089191207159003,
+      "loss": 0.0239,
+      "step": 9277
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020890129292769097,
+      "loss": 0.0017,
+      "step": 9278
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020888346415571884,
+      "loss": 0.0015,
+      "step": 9279
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002088656344002817,
+      "loss": 0.0157,
+      "step": 9280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020884780366167734,
+      "loss": 0.0055,
+      "step": 9281
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020882997194020362,
+      "loss": 0.017,
+      "step": 9282
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002088121392361583,
+      "loss": 0.0221,
+      "step": 9283
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020879430554983922,
+      "loss": 0.0175,
+      "step": 9284
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002087764708815443,
+      "loss": 0.0208,
+      "step": 9285
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002087586352315713,
+      "loss": 0.0104,
+      "step": 9286
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020874079860021822,
+      "loss": 0.0112,
+      "step": 9287
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020872296098778295,
+      "loss": 0.0122,
+      "step": 9288
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020870512239456337,
+      "loss": 0.0102,
+      "step": 9289
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020868728282085738,
+      "loss": 0.0112,
+      "step": 9290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020866944226696302,
+      "loss": 0.0339,
+      "step": 9291
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020865160073317823,
+      "loss": 0.0242,
+      "step": 9292
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000208633758219801,
+      "loss": 0.0041,
+      "step": 9293
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002086159147271293,
+      "loss": 0.0261,
+      "step": 9294
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002085980702554612,
+      "loss": 0.018,
+      "step": 9295
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002085802248050947,
+      "loss": 0.0145,
+      "step": 9296
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002085623783763279,
+      "loss": 0.0065,
+      "step": 9297
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020854453096945882,
+      "loss": 0.0109,
+      "step": 9298
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020852668258478557,
+      "loss": 0.0243,
+      "step": 9299
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020850883322260624,
+      "loss": 0.0172,
+      "step": 9300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002084909828832189,
+      "loss": 0.0096,
+      "step": 9301
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002084731315669218,
+      "loss": 0.0119,
+      "step": 9302
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000208455279274013,
+      "loss": 0.0187,
+      "step": 9303
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002084374260047907,
+      "loss": 0.0091,
+      "step": 9304
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002084195717595531,
+      "loss": 0.0141,
+      "step": 9305
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020840171653859838,
+      "loss": 0.0262,
+      "step": 9306
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020838386034222474,
+      "loss": 0.0277,
+      "step": 9307
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002083660031707304,
+      "loss": 0.0048,
+      "step": 9308
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020834814502441364,
+      "loss": 0.0253,
+      "step": 9309
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020833028590357274,
+      "loss": 0.0189,
+      "step": 9310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020831242580850593,
+      "loss": 0.0334,
+      "step": 9311
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020829456473951154,
+      "loss": 0.0206,
+      "step": 9312
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020827670269688787,
+      "loss": 0.014,
+      "step": 9313
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082588396809333,
+      "loss": 0.0177,
+      "step": 9314
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020824097569194613,
+      "loss": 0.0144,
+      "step": 9315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020822311073022474,
+      "loss": 0.0078,
+      "step": 9316
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082052447960675,
+      "loss": 0.0032,
+      "step": 9317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020818737788977275,
+      "loss": 0.0057,
+      "step": 9318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020816951001163895,
+      "loss": 0.0133,
+      "step": 9319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002081516411619646,
+      "loss": 0.0171,
+      "step": 9320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020813377134104802,
+      "loss": 0.0079,
+      "step": 9321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002081159005491877,
+      "loss": 0.0097,
+      "step": 9322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020809802878668217,
+      "loss": 0.0047,
+      "step": 9323
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002080801560538299,
+      "loss": 0.0063,
+      "step": 9324
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020806228235092944,
+      "loss": 0.0362,
+      "step": 9325
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020804440767827919,
+      "loss": 0.0058,
+      "step": 9326
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020802653203617777,
+      "loss": 0.0112,
+      "step": 9327
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020800865542492373,
+      "loss": 0.0068,
+      "step": 9328
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002079907778448157,
+      "loss": 0.0142,
+      "step": 9329
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020797289929615217,
+      "loss": 0.0048,
+      "step": 9330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002079550197792318,
+      "loss": 0.0035,
+      "step": 9331
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002079371392943532,
+      "loss": 0.0065,
+      "step": 9332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000207919257841815,
+      "loss": 0.0062,
+      "step": 9333
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020790137542191589,
+      "loss": 0.0122,
+      "step": 9334
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020788349203495446,
+      "loss": 0.0281,
+      "step": 9335
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002078656076812295,
+      "loss": 0.0107,
+      "step": 9336
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020784772236103963,
+      "loss": 0.0127,
+      "step": 9337
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020782983607468367,
+      "loss": 0.0045,
+      "step": 9338
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002078119488224602,
+      "loss": 0.0105,
+      "step": 9339
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020779406060466812,
+      "loss": 0.0067,
+      "step": 9340
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002077761714216061,
+      "loss": 0.0322,
+      "step": 9341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020775828127357298,
+      "loss": 0.0082,
+      "step": 9342
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002077403901608675,
+      "loss": 0.0135,
+      "step": 9343
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020772249808378854,
+      "loss": 0.0074,
+      "step": 9344
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002077046050426349,
+      "loss": 0.0148,
+      "step": 9345
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020768671103770542,
+      "loss": 0.0204,
+      "step": 9346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000207668816069299,
+      "loss": 0.0253,
+      "step": 9347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020765092013771448,
+      "loss": 0.0022,
+      "step": 9348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020763302324325078,
+      "loss": 0.0095,
+      "step": 9349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020761512538620676,
+      "loss": 0.0079,
+      "step": 9350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002075972265668814,
+      "loss": 0.0383,
+      "step": 9351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020757932678557367,
+      "loss": 0.0173,
+      "step": 9352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020756142604258247,
+      "loss": 0.0104,
+      "step": 9353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020754352433820678,
+      "loss": 0.0117,
+      "step": 9354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020752562167274563,
+      "loss": 0.0085,
+      "step": 9355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020750771804649802,
+      "loss": 0.032,
+      "step": 9356
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207489813459763,
+      "loss": 0.0186,
+      "step": 9357
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020747190791283953,
+      "loss": 0.0175,
+      "step": 9358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002074540014060267,
+      "loss": 0.0325,
+      "step": 9359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020743609393962363,
+      "loss": 0.0191,
+      "step": 9360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020741818551392937,
+      "loss": 0.0123,
+      "step": 9361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020740027612924303,
+      "loss": 0.0475,
+      "step": 9362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020738236578586366,
+      "loss": 0.0389,
+      "step": 9363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002073644544840905,
+      "loss": 0.01,
+      "step": 9364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020734654222422268,
+      "loss": 0.0204,
+      "step": 9365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002073286290065594,
+      "loss": 0.0133,
+      "step": 9366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020731071483139977,
+      "loss": 0.0074,
+      "step": 9367
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207292799699043,
+      "loss": 0.0052,
+      "step": 9368
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020727488360978833,
+      "loss": 0.0316,
+      "step": 9369
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207256966563935,
+      "loss": 0.0246,
+      "step": 9370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020723904856178227,
+      "loss": 0.0372,
+      "step": 9371
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020722112960362937,
+      "loss": 0.0028,
+      "step": 9372
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020720320968977557,
+      "loss": 0.0037,
+      "step": 9373
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071852888205202,
+      "loss": 0.0069,
+      "step": 9374
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071673669961625,
+      "loss": 0.0113,
+      "step": 9375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207149444217002,
+      "loss": 0.0139,
+      "step": 9376
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071315204833378,
+      "loss": 0.0112,
+      "step": 9377
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071135957954694,
+      "loss": 0.0056,
+      "step": 9378
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020709567015369612,
+      "loss": 0.0085,
+      "step": 9379
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002070777435583174,
+      "loss": 0.0037,
+      "step": 9380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002070598160096326,
+      "loss": 0.0051,
+      "step": 9381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020704188750794117,
+      "loss": 0.0079,
+      "step": 9382
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020702395805354256,
+      "loss": 0.012,
+      "step": 9383
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020700602764673617,
+      "loss": 0.0048,
+      "step": 9384
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020698809628782154,
+      "loss": 0.0136,
+      "step": 9385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020697016397709813,
+      "loss": 0.0076,
+      "step": 9386
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020695223071486542,
+      "loss": 0.0166,
+      "step": 9387
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020693429650142298,
+      "loss": 0.0067,
+      "step": 9388
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002069163613370703,
+      "loss": 0.0296,
+      "step": 9389
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000206898425222107,
+      "loss": 0.0093,
+      "step": 9390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020688048815683253,
+      "loss": 0.02,
+      "step": 9391
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002068625501415466,
+      "loss": 0.016,
+      "step": 9392
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002068446111765487,
+      "loss": 0.0064,
+      "step": 9393
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020682667126213847,
+      "loss": 0.0103,
+      "step": 9394
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002068087303986156,
+      "loss": 0.008,
+      "step": 9395
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020679078858627965,
+      "loss": 0.0401,
+      "step": 9396
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002067728458254304,
+      "loss": 0.0089,
+      "step": 9397
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020675490211636744,
+      "loss": 0.0165,
+      "step": 9398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020673695745939048,
+      "loss": 0.0076,
+      "step": 9399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002067190118547992,
+      "loss": 0.0134,
+      "step": 9400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002067010653028934,
+      "loss": 0.0065,
+      "step": 9401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020668311780397273,
+      "loss": 0.0251,
+      "step": 9402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020666516935833706,
+      "loss": 0.0135,
+      "step": 9403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020664721996628606,
+      "loss": 0.0125,
+      "step": 9404
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020662926962811953,
+      "loss": 0.0176,
+      "step": 9405
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020661131834413734,
+      "loss": 0.0219,
+      "step": 9406
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020659336611463923,
+      "loss": 0.0149,
+      "step": 9407
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020657541293992517,
+      "loss": 0.024,
+      "step": 9408
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020655745882029486,
+      "loss": 0.0178,
+      "step": 9409
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002065395037560482,
+      "loss": 0.0141,
+      "step": 9410
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020652154774748513,
+      "loss": 0.0078,
+      "step": 9411
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020650359079490554,
+      "loss": 0.0024,
+      "step": 9412
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020648563289860932,
+      "loss": 0.0183,
+      "step": 9413
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002064676740588964,
+      "loss": 0.0118,
+      "step": 9414
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020644971427606672,
+      "loss": 0.0107,
+      "step": 9415
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020643175355042023,
+      "loss": 0.0084,
+      "step": 9416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000206413791882257,
+      "loss": 0.0236,
+      "step": 9417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002063958292718769,
+      "loss": 0.0099,
+      "step": 9418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020637786571958003,
+      "loss": 0.0077,
+      "step": 9419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020635990122566637,
+      "loss": 0.0173,
+      "step": 9420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020634193579043594,
+      "loss": 0.0084,
+      "step": 9421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002063239694141889,
+      "loss": 0.0101,
+      "step": 9422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002063060020972252,
+      "loss": 0.0086,
+      "step": 9423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020628803383984502,
+      "loss": 0.0009,
+      "step": 9424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020627006464234837,
+      "loss": 0.0113,
+      "step": 9425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020625209450503545,
+      "loss": 0.0129,
+      "step": 9426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002062341234282064,
+      "loss": 0.0064,
+      "step": 9427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002062161514121613,
+      "loss": 0.0139,
+      "step": 9428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020619817845720035,
+      "loss": 0.0123,
+      "step": 9429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020618020456362375,
+      "loss": 0.0016,
+      "step": 9430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002061622297317317,
+      "loss": 0.0099,
+      "step": 9431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002061442539618244,
+      "loss": 0.005,
+      "step": 9432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020612627725420207,
+      "loss": 0.0077,
+      "step": 9433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020610829960916496,
+      "loss": 0.0202,
+      "step": 9434
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020609032102701335,
+      "loss": 0.0258,
+      "step": 9435
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020607234150804747,
+      "loss": 0.0093,
+      "step": 9436
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020605436105256767,
+      "loss": 0.0157,
+      "step": 9437
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020603637966087418,
+      "loss": 0.0055,
+      "step": 9438
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020601839733326742,
+      "loss": 0.0057,
+      "step": 9439
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020600041407004772,
+      "loss": 0.0124,
+      "step": 9440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020598242987151534,
+      "loss": 0.0056,
+      "step": 9441
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020596444473797073,
+      "loss": 0.0181,
+      "step": 9442
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020594645866971423,
+      "loss": 0.0213,
+      "step": 9443
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002059284716670463,
+      "loss": 0.0028,
+      "step": 9444
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002059104837302673,
+      "loss": 0.0164,
+      "step": 9445
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020589249485967765,
+      "loss": 0.0072,
+      "step": 9446
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020587450505557789,
+      "loss": 0.0162,
+      "step": 9447
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020585651431826837,
+      "loss": 0.0284,
+      "step": 9448
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020583852264804967,
+      "loss": 0.0306,
+      "step": 9449
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002058205300452222,
+      "loss": 0.0409,
+      "step": 9450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020580253651008654,
+      "loss": 0.0044,
+      "step": 9451
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002057845420429432,
+      "loss": 0.0157,
+      "step": 9452
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020576654664409267,
+      "loss": 0.0097,
+      "step": 9453
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020574855031383555,
+      "loss": 0.0046,
+      "step": 9454
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020573055305247241,
+      "loss": 0.0188,
+      "step": 9455
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020571255486030385,
+      "loss": 0.0323,
+      "step": 9456
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020569455573763045,
+      "loss": 0.0061,
+      "step": 9457
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002056765556847528,
+      "loss": 0.0137,
+      "step": 9458
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002056585547019716,
+      "loss": 0.0138,
+      "step": 9459
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020564055278958753,
+      "loss": 0.0066,
+      "step": 9460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002056225499479011,
+      "loss": 0.0129,
+      "step": 9461
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020560454617721318,
+      "loss": 0.004,
+      "step": 9462
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020558654147782434,
+      "loss": 0.0192,
+      "step": 9463
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020556853585003534,
+      "loss": 0.04,
+      "step": 9464
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002055505292941469,
+      "loss": 0.0155,
+      "step": 9465
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020553252181045972,
+      "loss": 0.0217,
+      "step": 9466
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020551451339927466,
+      "loss": 0.0076,
+      "step": 9467
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002054965040608924,
+      "loss": 0.0171,
+      "step": 9468
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020547849379561376,
+      "loss": 0.0044,
+      "step": 9469
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020546048260373958,
+      "loss": 0.0077,
+      "step": 9470
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020544247048557058,
+      "loss": 0.0058,
+      "step": 9471
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020542445744140775,
+      "loss": 0.0054,
+      "step": 9472
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020540644347155187,
+      "loss": 0.0056,
+      "step": 9473
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020538842857630374,
+      "loss": 0.0193,
+      "step": 9474
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020537041275596433,
+      "loss": 0.0406,
+      "step": 9475
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020535239601083448,
+      "loss": 0.014,
+      "step": 9476
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020533437834121512,
+      "loss": 0.0402,
+      "step": 9477
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020531635974740722,
+      "loss": 0.0164,
+      "step": 9478
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002052983402297117,
+      "loss": 0.0059,
+      "step": 9479
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002052803197884295,
+      "loss": 0.0209,
+      "step": 9480
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020526229842386163,
+      "loss": 0.0045,
+      "step": 9481
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020524427613630907,
+      "loss": 0.0154,
+      "step": 9482
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020522625292607278,
+      "loss": 0.016,
+      "step": 9483
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020520822879345385,
+      "loss": 0.0051,
+      "step": 9484
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020519020373875328,
+      "loss": 0.0195,
+      "step": 9485
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020517217776227207,
+      "loss": 0.0038,
+      "step": 9486
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020515415086431145,
+      "loss": 0.0011,
+      "step": 9487
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020513612304517233,
+      "loss": 0.0244,
+      "step": 9488
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020511809430515587,
+      "loss": 0.0084,
+      "step": 9489
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020510006464456322,
+      "loss": 0.0185,
+      "step": 9490
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020508203406369553,
+      "loss": 0.0062,
+      "step": 9491
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020506400256285385,
+      "loss": 0.0059,
+      "step": 9492
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020504597014233935,
+      "loss": 0.0084,
+      "step": 9493
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020502793680245328,
+      "loss": 0.0322,
+      "step": 9494
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020500990254349678,
+      "loss": 0.0687,
+      "step": 9495
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002049918673657711,
+      "loss": 0.0147,
+      "step": 9496
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020497383126957743,
+      "loss": 0.0222,
+      "step": 9497
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020495579425521695,
+      "loss": 0.0194,
+      "step": 9498
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020493775632299098,
+      "loss": 0.0079,
+      "step": 9499
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002049197174732008,
+      "loss": 0.0038,
+      "step": 9500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002049016777061477,
+      "loss": 0.0141,
+      "step": 9501
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020488363702213293,
+      "loss": 0.0126,
+      "step": 9502
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002048655954214578,
+      "loss": 0.0097,
+      "step": 9503
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020484755290442366,
+      "loss": 0.0064,
+      "step": 9504
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002048295094713319,
+      "loss": 0.0169,
+      "step": 9505
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002048114651224838,
+      "loss": 0.016,
+      "step": 9506
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020479341985818076,
+      "loss": 0.009,
+      "step": 9507
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020477537367872417,
+      "loss": 0.0313,
+      "step": 9508
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020475732658441548,
+      "loss": 0.0244,
+      "step": 9509
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020473927857555602,
+      "loss": 0.0066,
+      "step": 9510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020472122965244736,
+      "loss": 0.0198,
+      "step": 9511
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020470317981539078,
+      "loss": 0.0032,
+      "step": 9512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002046851290646879,
+      "loss": 0.0507,
+      "step": 9513
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020466707740064013,
+      "loss": 0.0077,
+      "step": 9514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020464902482354898,
+      "loss": 0.0061,
+      "step": 9515
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020463097133371596,
+      "loss": 0.0199,
+      "step": 9516
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020461291693144254,
+      "loss": 0.0164,
+      "step": 9517
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020459486161703038,
+      "loss": 0.0244,
+      "step": 9518
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020457680539078094,
+      "loss": 0.0071,
+      "step": 9519
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020455874825299586,
+      "loss": 0.0047,
+      "step": 9520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020454069020397667,
+      "loss": 0.0065,
+      "step": 9521
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000204522631244025,
+      "loss": 0.0162,
+      "step": 9522
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002045045713734425,
+      "loss": 0.0204,
+      "step": 9523
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020448651059253077,
+      "loss": 0.0351,
+      "step": 9524
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020446844890159146,
+      "loss": 0.0098,
+      "step": 9525
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002044503863009262,
+      "loss": 0.0026,
+      "step": 9526
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020443232279083673,
+      "loss": 0.017,
+      "step": 9527
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020441425837162478,
+      "loss": 0.0197,
+      "step": 9528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002043961930435919,
+      "loss": 0.0156,
+      "step": 9529
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020437812680703995,
+      "loss": 0.0175,
+      "step": 9530
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020436005966227065,
+      "loss": 0.0184,
+      "step": 9531
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002043419916095857,
+      "loss": 0.0123,
+      "step": 9532
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020432392264928698,
+      "loss": 0.0086,
+      "step": 9533
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020430585278167617,
+      "loss": 0.0244,
+      "step": 9534
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020428778200705513,
+      "loss": 0.0184,
+      "step": 9535
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020426971032572563,
+      "loss": 0.0195,
+      "step": 9536
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020425163773798955,
+      "loss": 0.0098,
+      "step": 9537
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020423356424414873,
+      "loss": 0.012,
+      "step": 9538
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020421548984450496,
+      "loss": 0.0054,
+      "step": 9539
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020419741453936023,
+      "loss": 0.0143,
+      "step": 9540
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020417933832901628,
+      "loss": 0.0319,
+      "step": 9541
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002041612612137752,
+      "loss": 0.0161,
+      "step": 9542
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020414318319393878,
+      "loss": 0.0053,
+      "step": 9543
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000204125104269809,
+      "loss": 0.0035,
+      "step": 9544
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002041070244416878,
+      "loss": 0.0124,
+      "step": 9545
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020408894370987717,
+      "loss": 0.0219,
+      "step": 9546
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020407086207467913,
+      "loss": 0.0124,
+      "step": 9547
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020405277953639556,
+      "loss": 0.0082,
+      "step": 9548
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020403469609532856,
+      "loss": 0.0213,
+      "step": 9549
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020401661175178018,
+      "loss": 0.0237,
+      "step": 9550
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020399852650605235,
+      "loss": 0.0068,
+      "step": 9551
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002039804403584472,
+      "loss": 0.0034,
+      "step": 9552
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020396235330926682,
+      "loss": 0.0109,
+      "step": 9553
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020394426535881333,
+      "loss": 0.0045,
+      "step": 9554
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020392617650738873,
+      "loss": 0.0077,
+      "step": 9555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002039080867552952,
+      "loss": 0.0167,
+      "step": 9556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020388999610283486,
+      "loss": 0.0081,
+      "step": 9557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020387190455030984,
+      "loss": 0.0089,
+      "step": 9558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020385381209802236,
+      "loss": 0.0095,
+      "step": 9559
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020383571874627455,
+      "loss": 0.024,
+      "step": 9560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020381762449536857,
+      "loss": 0.0247,
+      "step": 9561
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020379952934560671,
+      "loss": 0.0124,
+      "step": 9562
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002037814332972912,
+      "loss": 0.0145,
+      "step": 9563
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002037633363507242,
+      "loss": 0.0063,
+      "step": 9564
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000203745238506208,
+      "loss": 0.0302,
+      "step": 9565
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020372713976404486,
+      "loss": 0.0126,
+      "step": 9566
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020370904012453705,
+      "loss": 0.0084,
+      "step": 9567
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002036909395879869,
+      "loss": 0.013,
+      "step": 9568
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002036728381546967,
+      "loss": 0.0172,
+      "step": 9569
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020365473582496885,
+      "loss": 0.0094,
+      "step": 9570
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020363663259910555,
+      "loss": 0.006,
+      "step": 9571
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020361852847740927,
+      "loss": 0.0159,
+      "step": 9572
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020360042346018233,
+      "loss": 0.008,
+      "step": 9573
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020358231754772716,
+      "loss": 0.0079,
+      "step": 9574
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020356421074034617,
+      "loss": 0.0102,
+      "step": 9575
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002035461030383417,
+      "loss": 0.0068,
+      "step": 9576
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002035279944420163,
+      "loss": 0.0145,
+      "step": 9577
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002035098849516723,
+      "loss": 0.0216,
+      "step": 9578
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020349177456761223,
+      "loss": 0.0075,
+      "step": 9579
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020347366329013855,
+      "loss": 0.0074,
+      "step": 9580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020345555111955378,
+      "loss": 0.0151,
+      "step": 9581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020343743805616035,
+      "loss": 0.0288,
+      "step": 9582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020341932410026086,
+      "loss": 0.0057,
+      "step": 9583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020340120925215784,
+      "loss": 0.0297,
+      "step": 9584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020338309351215376,
+      "loss": 0.0082,
+      "step": 9585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020336497688055132,
+      "loss": 0.0067,
+      "step": 9586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000203346859357653,
+      "loss": 0.0029,
+      "step": 9587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020332874094376147,
+      "loss": 0.0224,
+      "step": 9588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020331062163917924,
+      "loss": 0.0174,
+      "step": 9589
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020329250144420904,
+      "loss": 0.0184,
+      "step": 9590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020327438035915344,
+      "loss": 0.0177,
+      "step": 9591
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020325625838431516,
+      "loss": 0.005,
+      "step": 9592
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002032381355199968,
+      "loss": 0.0019,
+      "step": 9593
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002032200117665011,
+      "loss": 0.0098,
+      "step": 9594
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020320188712413074,
+      "loss": 0.0123,
+      "step": 9595
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020318376159318845,
+      "loss": 0.0292,
+      "step": 9596
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020316563517397694,
+      "loss": 0.0083,
+      "step": 9597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020314750786679897,
+      "loss": 0.014,
+      "step": 9598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020312937967195734,
+      "loss": 0.0035,
+      "step": 9599
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002031112505897547,
+      "loss": 0.0118,
+      "step": 9600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020309312062049396,
+      "loss": 0.0102,
+      "step": 9601
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020307498976447787,
+      "loss": 0.0058,
+      "step": 9602
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020305685802200928,
+      "loss": 0.0073,
+      "step": 9603
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020303872539339097,
+      "loss": 0.0073,
+      "step": 9604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020302059187892588,
+      "loss": 0.0139,
+      "step": 9605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020300245747891681,
+      "loss": 0.0117,
+      "step": 9606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020298432219366665,
+      "loss": 0.0066,
+      "step": 9607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002029661860234783,
+      "loss": 0.0162,
+      "step": 9608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020294804896865467,
+      "loss": 0.0183,
+      "step": 9609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020292991102949866,
+      "loss": 0.0089,
+      "step": 9610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020291177220631317,
+      "loss": 0.0087,
+      "step": 9611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002028936324994013,
+      "loss": 0.0029,
+      "step": 9612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020287549190906587,
+      "loss": 0.0109,
+      "step": 9613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002028573504356099,
+      "loss": 0.0081,
+      "step": 9614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020283920807933644,
+      "loss": 0.01,
+      "step": 9615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020282106484054846,
+      "loss": 0.0183,
+      "step": 9616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020280292071954897,
+      "loss": 0.0066,
+      "step": 9617
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020278477571664106,
+      "loss": 0.022,
+      "step": 9618
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020276662983212773,
+      "loss": 0.0156,
+      "step": 9619
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002027484830663121,
+      "loss": 0.0205,
+      "step": 9620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020273033541949725,
+      "loss": 0.0044,
+      "step": 9621
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020271218689198623,
+      "loss": 0.0068,
+      "step": 9622
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020269403748408217,
+      "loss": 0.0241,
+      "step": 9623
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020267588719608825,
+      "loss": 0.0028,
+      "step": 9624
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002026577360283076,
+      "loss": 0.0208,
+      "step": 9625
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002026395839810433,
+      "loss": 0.0189,
+      "step": 9626
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020262143105459861,
+      "loss": 0.0172,
+      "step": 9627
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020260327724927672,
+      "loss": 0.0077,
+      "step": 9628
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020258512256538078,
+      "loss": 0.008,
+      "step": 9629
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020256696700321402,
+      "loss": 0.0132,
+      "step": 9630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002025488105630797,
+      "loss": 0.0145,
+      "step": 9631
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020253065324528106,
+      "loss": 0.0108,
+      "step": 9632
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020251249505012133,
+      "loss": 0.0099,
+      "step": 9633
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002024943359779038,
+      "loss": 0.0027,
+      "step": 9634
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002024761760289318,
+      "loss": 0.0409,
+      "step": 9635
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020245801520350857,
+      "loss": 0.0276,
+      "step": 9636
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020243985350193748,
+      "loss": 0.0085,
+      "step": 9637
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020242169092452188,
+      "loss": 0.0164,
+      "step": 9638
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020240352747156505,
+      "loss": 0.0227,
+      "step": 9639
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002023853631433704,
+      "loss": 0.0368,
+      "step": 9640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020236719794024133,
+      "loss": 0.0156,
+      "step": 9641
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020234903186248118,
+      "loss": 0.0094,
+      "step": 9642
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020233086491039338,
+      "loss": 0.0178,
+      "step": 9643
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020231269708428137,
+      "loss": 0.0045,
+      "step": 9644
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020229452838444855,
+      "loss": 0.0259,
+      "step": 9645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020227635881119838,
+      "loss": 0.0041,
+      "step": 9646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020225818836483436,
+      "loss": 0.0102,
+      "step": 9647
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020224001704565993,
+      "loss": 0.0217,
+      "step": 9648
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020222184485397867,
+      "loss": 0.0421,
+      "step": 9649
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020220367179009401,
+      "loss": 0.0068,
+      "step": 9650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020218549785430944,
+      "loss": 0.0203,
+      "step": 9651
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020216732304692854,
+      "loss": 0.0095,
+      "step": 9652
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020214914736825496,
+      "loss": 0.0195,
+      "step": 9653
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002021309708185921,
+      "loss": 0.0103,
+      "step": 9654
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020211279339824362,
+      "loss": 0.0076,
+      "step": 9655
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020209461510751312,
+      "loss": 0.0262,
+      "step": 9656
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020207643594670425,
+      "loss": 0.0369,
+      "step": 9657
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020205825591612062,
+      "loss": 0.0092,
+      "step": 9658
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002020400750160658,
+      "loss": 0.0077,
+      "step": 9659
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020202189324684353,
+      "loss": 0.0394,
+      "step": 9660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002020037106087574,
+      "loss": 0.0051,
+      "step": 9661
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002019855271021112,
+      "loss": 0.0269,
+      "step": 9662
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020196734272720854,
+      "loss": 0.0149,
+      "step": 9663
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020194915748435313,
+      "loss": 0.0123,
+      "step": 9664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020193097137384875,
+      "loss": 0.0247,
+      "step": 9665
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020191278439599908,
+      "loss": 0.02,
+      "step": 9666
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000201894596551108,
+      "loss": 0.0102,
+      "step": 9667
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020187640783947915,
+      "loss": 0.0078,
+      "step": 9668
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002018582182614164,
+      "loss": 0.0066,
+      "step": 9669
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002018400278172235,
+      "loss": 0.0128,
+      "step": 9670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020182183650720425,
+      "loss": 0.0122,
+      "step": 9671
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002018036443316626,
+      "loss": 0.005,
+      "step": 9672
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002017854512909022,
+      "loss": 0.0137,
+      "step": 9673
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020176725738522706,
+      "loss": 0.0125,
+      "step": 9674
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020174906261494096,
+      "loss": 0.0117,
+      "step": 9675
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020173086698034787,
+      "loss": 0.0222,
+      "step": 9676
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020171267048175166,
+      "loss": 0.008,
+      "step": 9677
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002016944731194562,
+      "loss": 0.0152,
+      "step": 9678
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020167627489376546,
+      "loss": 0.0232,
+      "step": 9679
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020165807580498342,
+      "loss": 0.0057,
+      "step": 9680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000201639875853414,
+      "loss": 0.0119,
+      "step": 9681
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002016216750393612,
+      "loss": 0.0191,
+      "step": 9682
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020160347336312896,
+      "loss": 0.0115,
+      "step": 9683
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020158527082502127,
+      "loss": 0.0142,
+      "step": 9684
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020156706742534218,
+      "loss": 0.011,
+      "step": 9685
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020154886316439577,
+      "loss": 0.0284,
+      "step": 9686
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000201530658042486,
+      "loss": 0.0028,
+      "step": 9687
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020151245205991702,
+      "loss": 0.0164,
+      "step": 9688
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020149424521699282,
+      "loss": 0.0163,
+      "step": 9689
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020147603751401755,
+      "loss": 0.0186,
+      "step": 9690
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020145782895129524,
+      "loss": 0.0276,
+      "step": 9691
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020143961952913007,
+      "loss": 0.0079,
+      "step": 9692
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020142140924782616,
+      "loss": 0.0094,
+      "step": 9693
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020140319810768765,
+      "loss": 0.0419,
+      "step": 9694
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020138498610901872,
+      "loss": 0.0107,
+      "step": 9695
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002013667732521235,
+      "loss": 0.011,
+      "step": 9696
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002013485595373062,
+      "loss": 0.0311,
+      "step": 9697
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020133034496487107,
+      "loss": 0.0276,
+      "step": 9698
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020131212953512228,
+      "loss": 0.037,
+      "step": 9699
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020129391324836402,
+      "loss": 0.0109,
+      "step": 9700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002012756961049006,
+      "loss": 0.018,
+      "step": 9701
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020125747810503626,
+      "loss": 0.0219,
+      "step": 9702
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020123925924907533,
+      "loss": 0.0042,
+      "step": 9703
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020122103953732203,
+      "loss": 0.0091,
+      "step": 9704
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020120281897008068,
+      "loss": 0.0025,
+      "step": 9705
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002011845975476556,
+      "loss": 0.0161,
+      "step": 9706
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020116637527035105,
+      "loss": 0.0139,
+      "step": 9707
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020114815213847153,
+      "loss": 0.0063,
+      "step": 9708
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020112992815232134,
+      "loss": 0.0163,
+      "step": 9709
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020111170331220477,
+      "loss": 0.0057,
+      "step": 9710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020109347761842632,
+      "loss": 0.0175,
+      "step": 9711
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002010752510712903,
+      "loss": 0.0088,
+      "step": 9712
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020105702367110125,
+      "loss": 0.0031,
+      "step": 9713
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020103879541816347,
+      "loss": 0.0188,
+      "step": 9714
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002010205663127815,
+      "loss": 0.0035,
+      "step": 9715
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020100233635525976,
+      "loss": 0.0254,
+      "step": 9716
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020098410554590266,
+      "loss": 0.0187,
+      "step": 9717
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002009658738850148,
+      "loss": 0.0065,
+      "step": 9718
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020094764137290066,
+      "loss": 0.0037,
+      "step": 9719
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020092940800986475,
+      "loss": 0.0102,
+      "step": 9720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020091117379621157,
+      "loss": 0.0173,
+      "step": 9721
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020089293873224568,
+      "loss": 0.0031,
+      "step": 9722
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020087470281827168,
+      "loss": 0.0068,
+      "step": 9723
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020085646605459408,
+      "loss": 0.0107,
+      "step": 9724
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002008382284415175,
+      "loss": 0.0144,
+      "step": 9725
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020081998997934655,
+      "loss": 0.004,
+      "step": 9726
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002008017506683858,
+      "loss": 0.0054,
+      "step": 9727
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020078351050893995,
+      "loss": 0.0096,
+      "step": 9728
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020076526950131362,
+      "loss": 0.0105,
+      "step": 9729
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020074702764581148,
+      "loss": 0.0131,
+      "step": 9730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020072878494273812,
+      "loss": 0.0079,
+      "step": 9731
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020071054139239838,
+      "loss": 0.0106,
+      "step": 9732
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020069229699509682,
+      "loss": 0.0112,
+      "step": 9733
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020067405175113823,
+      "loss": 0.0132,
+      "step": 9734
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020065580566082733,
+      "loss": 0.0353,
+      "step": 9735
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020063755872446888,
+      "loss": 0.0087,
+      "step": 9736
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020061931094236756,
+      "loss": 0.0069,
+      "step": 9737
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020060106231482817,
+      "loss": 0.0077,
+      "step": 9738
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020058281284215558,
+      "loss": 0.025,
+      "step": 9739
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002005645625246545,
+      "loss": 0.0127,
+      "step": 9740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020054631136262984,
+      "loss": 0.0153,
+      "step": 9741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020052805935638632,
+      "loss": 0.0079,
+      "step": 9742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002005098065062288,
+      "loss": 0.0204,
+      "step": 9743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020049155281246222,
+      "loss": 0.0316,
+      "step": 9744
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020047329827539137,
+      "loss": 0.0145,
+      "step": 9745
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002004550428953212,
+      "loss": 0.0213,
+      "step": 9746
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020043678667255657,
+      "loss": 0.0149,
+      "step": 9747
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020041852960740234,
+      "loss": 0.0098,
+      "step": 9748
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002004002717001635,
+      "loss": 0.0058,
+      "step": 9749
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020038201295114507,
+      "loss": 0.026,
+      "step": 9750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020036375336065184,
+      "loss": 0.0121,
+      "step": 9751
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020034549292898888,
+      "loss": 0.0124,
+      "step": 9752
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020032723165646117,
+      "loss": 0.006,
+      "step": 9753
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020030896954337368,
+      "loss": 0.0072,
+      "step": 9754
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020029070659003146,
+      "loss": 0.021,
+      "step": 9755
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020027244279673947,
+      "loss": 0.0237,
+      "step": 9756
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020025417816380277,
+      "loss": 0.0062,
+      "step": 9757
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002002359126915265,
+      "loss": 0.0049,
+      "step": 9758
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002002176463802156,
+      "loss": 0.012,
+      "step": 9759
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020019937923017523,
+      "loss": 0.0056,
+      "step": 9760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002001811112417105,
+      "loss": 0.0031,
+      "step": 9761
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020016284241512647,
+      "loss": 0.0134,
+      "step": 9762
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020014457275072825,
+      "loss": 0.0213,
+      "step": 9763
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002001263022488211,
+      "loss": 0.012,
+      "step": 9764
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020010803090971002,
+      "loss": 0.0096,
+      "step": 9765
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020008975873370022,
+      "loss": 0.0093,
+      "step": 9766
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020007148572109696,
+      "loss": 0.0056,
+      "step": 9767
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002000532118722053,
+      "loss": 0.0138,
+      "step": 9768
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020003493718733054,
+      "loss": 0.0024,
+      "step": 9769
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002000166616667779,
+      "loss": 0.0075,
+      "step": 9770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001999983853108526,
+      "loss": 0.0123,
+      "step": 9771
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019998010811985987,
+      "loss": 0.0101,
+      "step": 9772
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019996183009410496,
+      "loss": 0.0151,
+      "step": 9773
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019994355123389325,
+      "loss": 0.0087,
+      "step": 9774
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019992527153952995,
+      "loss": 0.0143,
+      "step": 9775
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019990699101132034,
+      "loss": 0.0031,
+      "step": 9776
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019988870964956977,
+      "loss": 0.0146,
+      "step": 9777
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001998704274545836,
+      "loss": 0.0063,
+      "step": 9778
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019985214442666712,
+      "loss": 0.0152,
+      "step": 9779
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019983386056612575,
+      "loss": 0.0177,
+      "step": 9780
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019981557587326484,
+      "loss": 0.0098,
+      "step": 9781
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019979729034838977,
+      "loss": 0.0115,
+      "step": 9782
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019977900399180596,
+      "loss": 0.0222,
+      "step": 9783
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019976071680381885,
+      "loss": 0.039,
+      "step": 9784
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001997424287847338,
+      "loss": 0.0082,
+      "step": 9785
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001997241399348563,
+      "loss": 0.0129,
+      "step": 9786
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019970585025449184,
+      "loss": 0.0095,
+      "step": 9787
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019968755974394582,
+      "loss": 0.0121,
+      "step": 9788
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019966926840352377,
+      "loss": 0.011,
+      "step": 9789
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019965097623353116,
+      "loss": 0.0057,
+      "step": 9790
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019963268323427348,
+      "loss": 0.002,
+      "step": 9791
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001996143894060564,
+      "loss": 0.0188,
+      "step": 9792
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019959609474918532,
+      "loss": 0.0131,
+      "step": 9793
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019957779926396585,
+      "loss": 0.0183,
+      "step": 9794
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019955950295070352,
+      "loss": 0.0088,
+      "step": 9795
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019954120580970393,
+      "loss": 0.0145,
+      "step": 9796
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019952290784127277,
+      "loss": 0.0298,
+      "step": 9797
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019950460904571548,
+      "loss": 0.0147,
+      "step": 9798
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019948630942333783,
+      "loss": 0.0079,
+      "step": 9799
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019946800897444538,
+      "loss": 0.0376,
+      "step": 9800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001994497076993438,
+      "loss": 0.0064,
+      "step": 9801
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001994314055983387,
+      "loss": 0.0053,
+      "step": 9802
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001994131026717359,
+      "loss": 0.0106,
+      "step": 9803
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000199394798919841,
+      "loss": 0.0244,
+      "step": 9804
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019937649434295968,
+      "loss": 0.0287,
+      "step": 9805
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019935818894139775,
+      "loss": 0.0154,
+      "step": 9806
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019933988271546088,
+      "loss": 0.0176,
+      "step": 9807
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019932157566545478,
+      "loss": 0.0351,
+      "step": 9808
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019930326779168532,
+      "loss": 0.0028,
+      "step": 9809
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001992849590944582,
+      "loss": 0.0198,
+      "step": 9810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019926664957407926,
+      "loss": 0.0134,
+      "step": 9811
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019924833923085422,
+      "loss": 0.0288,
+      "step": 9812
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019923002806508902,
+      "loss": 0.0131,
+      "step": 9813
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019921171607708936,
+      "loss": 0.0186,
+      "step": 9814
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019919340326716122,
+      "loss": 0.0079,
+      "step": 9815
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019917508963561033,
+      "loss": 0.0016,
+      "step": 9816
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001991567751827426,
+      "loss": 0.0175,
+      "step": 9817
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019913845990886397,
+      "loss": 0.0288,
+      "step": 9818
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001991201438142803,
+      "loss": 0.0212,
+      "step": 9819
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019910182689929755,
+      "loss": 0.0134,
+      "step": 9820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019908350916422152,
+      "loss": 0.009,
+      "step": 9821
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019906519060935827,
+      "loss": 0.0085,
+      "step": 9822
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019904687123501374,
+      "loss": 0.0039,
+      "step": 9823
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019902855104149387,
+      "loss": 0.0086,
+      "step": 9824
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019901023002910466,
+      "loss": 0.0097,
+      "step": 9825
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019899190819815208,
+      "loss": 0.0089,
+      "step": 9826
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019897358554894215,
+      "loss": 0.0078,
+      "step": 9827
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001989552620817809,
+      "loss": 0.002,
+      "step": 9828
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019893693779697436,
+      "loss": 0.0072,
+      "step": 9829
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019891861269482862,
+      "loss": 0.0067,
+      "step": 9830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019890028677564965,
+      "loss": 0.0045,
+      "step": 9831
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019888196003974357,
+      "loss": 0.0165,
+      "step": 9832
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019886363248741652,
+      "loss": 0.0012,
+      "step": 9833
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019884530411897464,
+      "loss": 0.0041,
+      "step": 9834
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001988269749347239,
+      "loss": 0.0078,
+      "step": 9835
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019880864493497053,
+      "loss": 0.0035,
+      "step": 9836
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019879031412002067,
+      "loss": 0.0158,
+      "step": 9837
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019877198249018045,
+      "loss": 0.0042,
+      "step": 9838
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019875365004575605,
+      "loss": 0.0062,
+      "step": 9839
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001987353167870537,
+      "loss": 0.0082,
+      "step": 9840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019871698271437953,
+      "loss": 0.0045,
+      "step": 9841
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019869864782803978,
+      "loss": 0.0111,
+      "step": 9842
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001986803121283407,
+      "loss": 0.0063,
+      "step": 9843
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019866197561558855,
+      "loss": 0.0165,
+      "step": 9844
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019864363829008949,
+      "loss": 0.0057,
+      "step": 9845
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001986253001521499,
+      "loss": 0.0038,
+      "step": 9846
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019860696120207596,
+      "loss": 0.017,
+      "step": 9847
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001985886214401741,
+      "loss": 0.0101,
+      "step": 9848
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019857028086675048,
+      "loss": 0.0106,
+      "step": 9849
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001985519394821115,
+      "loss": 0.0072,
+      "step": 9850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019853359728656348,
+      "loss": 0.0027,
+      "step": 9851
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019851525428041277,
+      "loss": 0.012,
+      "step": 9852
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019849691046396574,
+      "loss": 0.0011,
+      "step": 9853
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019847856583752875,
+      "loss": 0.0117,
+      "step": 9854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001984602204014082,
+      "loss": 0.0113,
+      "step": 9855
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019844187415591052,
+      "loss": 0.0139,
+      "step": 9856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001984235271013421,
+      "loss": 0.0052,
+      "step": 9857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019840517923800936,
+      "loss": 0.0075,
+      "step": 9858
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019838683056621877,
+      "loss": 0.0054,
+      "step": 9859
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019836848108627673,
+      "loss": 0.0034,
+      "step": 9860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001983501307984898,
+      "loss": 0.009,
+      "step": 9861
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019833177970316436,
+      "loss": 0.0023,
+      "step": 9862
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019831342780060702,
+      "loss": 0.0029,
+      "step": 9863
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001982950750911242,
+      "loss": 0.0044,
+      "step": 9864
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001982767215750225,
+      "loss": 0.0031,
+      "step": 9865
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001982583672526084,
+      "loss": 0.003,
+      "step": 9866
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019824001212418848,
+      "loss": 0.0031,
+      "step": 9867
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001982216561900693,
+      "loss": 0.0038,
+      "step": 9868
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019820329945055744,
+      "loss": 0.0102,
+      "step": 9869
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019818494190595946,
+      "loss": 0.023,
+      "step": 9870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019816658355658204,
+      "loss": 0.0148,
+      "step": 9871
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019814822440273174,
+      "loss": 0.0264,
+      "step": 9872
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019812986444471518,
+      "loss": 0.0011,
+      "step": 9873
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019811150368283904,
+      "loss": 0.0085,
+      "step": 9874
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019809314211741,
+      "loss": 0.0027,
+      "step": 9875
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001980747797487347,
+      "loss": 0.013,
+      "step": 9876
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019805641657711978,
+      "loss": 0.0059,
+      "step": 9877
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019803805260287203,
+      "loss": 0.0047,
+      "step": 9878
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001980196878262981,
+      "loss": 0.0078,
+      "step": 9879
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019800132224770482,
+      "loss": 0.0253,
+      "step": 9880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019798295586739877,
+      "loss": 0.0147,
+      "step": 9881
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019796458868568678,
+      "loss": 0.008,
+      "step": 9882
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019794622070287566,
+      "loss": 0.0111,
+      "step": 9883
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001979278519192721,
+      "loss": 0.0021,
+      "step": 9884
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000197909482335183,
+      "loss": 0.0203,
+      "step": 9885
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019789111195091504,
+      "loss": 0.0133,
+      "step": 9886
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019787274076677518,
+      "loss": 0.0052,
+      "step": 9887
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019785436878307016,
+      "loss": 0.0102,
+      "step": 9888
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019783599600010687,
+      "loss": 0.0162,
+      "step": 9889
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019781762241819208,
+      "loss": 0.016,
+      "step": 9890
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001977992480376328,
+      "loss": 0.0041,
+      "step": 9891
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001977808728587358,
+      "loss": 0.0078,
+      "step": 9892
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019776249688180806,
+      "loss": 0.0182,
+      "step": 9893
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019774412010715648,
+      "loss": 0.0017,
+      "step": 9894
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019772574253508794,
+      "loss": 0.0012,
+      "step": 9895
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019770736416590943,
+      "loss": 0.0057,
+      "step": 9896
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019768898499992788,
+      "loss": 0.017,
+      "step": 9897
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019767060503745026,
+      "loss": 0.0088,
+      "step": 9898
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019765222427878354,
+      "loss": 0.0078,
+      "step": 9899
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019763384272423475,
+      "loss": 0.0124,
+      "step": 9900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019761546037411087,
+      "loss": 0.0108,
+      "step": 9901
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019759707722871888,
+      "loss": 0.0008,
+      "step": 9902
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001975786932883659,
+      "loss": 0.0176,
+      "step": 9903
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019756030855335891,
+      "loss": 0.0072,
+      "step": 9904
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000197541923024005,
+      "loss": 0.0095,
+      "step": 9905
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019752353670061119,
+      "loss": 0.0041,
+      "step": 9906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019750514958348465,
+      "loss": 0.0046,
+      "step": 9907
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001974867616729325,
+      "loss": 0.005,
+      "step": 9908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001974683729692617,
+      "loss": 0.0081,
+      "step": 9909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001974499834727795,
+      "loss": 0.0061,
+      "step": 9910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000197431593183793,
+      "loss": 0.013,
+      "step": 9911
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001974132021026094,
+      "loss": 0.0035,
+      "step": 9912
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001973948102295358,
+      "loss": 0.0112,
+      "step": 9913
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019737641756487938,
+      "loss": 0.0081,
+      "step": 9914
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019735802410894733,
+      "loss": 0.0059,
+      "step": 9915
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019733962986204692,
+      "loss": 0.0088,
+      "step": 9916
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019732123482448535,
+      "loss": 0.0129,
+      "step": 9917
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019730283899656984,
+      "loss": 0.0317,
+      "step": 9918
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001972844423786076,
+      "loss": 0.0051,
+      "step": 9919
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019726604497090594,
+      "loss": 0.013,
+      "step": 9920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001972476467737721,
+      "loss": 0.0602,
+      "step": 9921
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019722924778751337,
+      "loss": 0.001,
+      "step": 9922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019721084801243706,
+      "loss": 0.0298,
+      "step": 9923
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019719244744885044,
+      "loss": 0.0096,
+      "step": 9924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019717404609706087,
+      "loss": 0.0073,
+      "step": 9925
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001971556439573757,
+      "loss": 0.0022,
+      "step": 9926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019713724103010227,
+      "loss": 0.0071,
+      "step": 9927
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019711883731554793,
+      "loss": 0.0049,
+      "step": 9928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019710043281402005,
+      "loss": 0.01,
+      "step": 9929
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019708202752582605,
+      "loss": 0.005,
+      "step": 9930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019706362145127332,
+      "loss": 0.0197,
+      "step": 9931
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019704521459066926,
+      "loss": 0.0041,
+      "step": 9932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019702680694432133,
+      "loss": 0.0047,
+      "step": 9933
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019700839851253697,
+      "loss": 0.0188,
+      "step": 9934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019698998929562356,
+      "loss": 0.0192,
+      "step": 9935
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019697157929388868,
+      "loss": 0.0171,
+      "step": 9936
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019695316850763974,
+      "loss": 0.0113,
+      "step": 9937
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019693475693718425,
+      "loss": 0.0098,
+      "step": 9938
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019691634458282971,
+      "loss": 0.0079,
+      "step": 9939
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019689793144488367,
+      "loss": 0.004,
+      "step": 9940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019687951752365366,
+      "loss": 0.0074,
+      "step": 9941
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019686110281944724,
+      "loss": 0.0124,
+      "step": 9942
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001968426873325719,
+      "loss": 0.0134,
+      "step": 9943
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019682427106333528,
+      "loss": 0.0115,
+      "step": 9944
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019680585401204493,
+      "loss": 0.0105,
+      "step": 9945
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001967874361790085,
+      "loss": 0.0121,
+      "step": 9946
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019676901756453353,
+      "loss": 0.0102,
+      "step": 9947
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019675059816892768,
+      "loss": 0.0068,
+      "step": 9948
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019673217799249864,
+      "loss": 0.0233,
+      "step": 9949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000196713757035554,
+      "loss": 0.0109,
+      "step": 9950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019669533529840142,
+      "loss": 0.0137,
+      "step": 9951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001966769127813486,
+      "loss": 0.0105,
+      "step": 9952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019665848948470323,
+      "loss": 0.0047,
+      "step": 9953
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019664006540877303,
+      "loss": 0.0183,
+      "step": 9954
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019662164055386566,
+      "loss": 0.0144,
+      "step": 9955
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019660321492028892,
+      "loss": 0.0052,
+      "step": 9956
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001965847885083505,
+      "loss": 0.0153,
+      "step": 9957
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019656636131835817,
+      "loss": 0.0035,
+      "step": 9958
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019654793335061976,
+      "loss": 0.0147,
+      "step": 9959
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019652950460544294,
+      "loss": 0.0127,
+      "step": 9960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019651107508313558,
+      "loss": 0.0108,
+      "step": 9961
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019649264478400547,
+      "loss": 0.0083,
+      "step": 9962
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019647421370836047,
+      "loss": 0.005,
+      "step": 9963
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019645578185650833,
+      "loss": 0.0064,
+      "step": 9964
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019643734922875696,
+      "loss": 0.0143,
+      "step": 9965
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019641891582541416,
+      "loss": 0.0145,
+      "step": 9966
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019640048164678789,
+      "loss": 0.0053,
+      "step": 9967
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019638204669318596,
+      "loss": 0.0039,
+      "step": 9968
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019636361096491632,
+      "loss": 0.0064,
+      "step": 9969
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019634517446228683,
+      "loss": 0.0117,
+      "step": 9970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019632673718560548,
+      "loss": 0.0021,
+      "step": 9971
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019630829913518016,
+      "loss": 0.0129,
+      "step": 9972
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019628986031131884,
+      "loss": 0.0098,
+      "step": 9973
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019627142071432945,
+      "loss": 0.0033,
+      "step": 9974
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019625298034452,
+      "loss": 0.0128,
+      "step": 9975
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019623453920219848,
+      "loss": 0.002,
+      "step": 9976
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019621609728767285,
+      "loss": 0.0115,
+      "step": 9977
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019619765460125122,
+      "loss": 0.0107,
+      "step": 9978
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019617921114324147,
+      "loss": 0.009,
+      "step": 9979
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019616076691395177,
+      "loss": 0.0069,
+      "step": 9980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019614232191369016,
+      "loss": 0.0019,
+      "step": 9981
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019612387614276465,
+      "loss": 0.0094,
+      "step": 9982
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019610542960148331,
+      "loss": 0.0095,
+      "step": 9983
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001960869822901543,
+      "loss": 0.0035,
+      "step": 9984
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019606853420908567,
+      "loss": 0.0044,
+      "step": 9985
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019605008535858562,
+      "loss": 0.008,
+      "step": 9986
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019603163573896212,
+      "loss": 0.0052,
+      "step": 9987
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019601318535052346,
+      "loss": 0.0095,
+      "step": 9988
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019599473419357774,
+      "loss": 0.0075,
+      "step": 9989
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019597628226843312,
+      "loss": 0.0049,
+      "step": 9990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019595782957539787,
+      "loss": 0.0191,
+      "step": 9991
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019593937611478006,
+      "loss": 0.0029,
+      "step": 9992
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019592092188688796,
+      "loss": 0.0217,
+      "step": 9993
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019590246689202978,
+      "loss": 0.0037,
+      "step": 9994
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019588401113051376,
+      "loss": 0.0087,
+      "step": 9995
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019586555460264816,
+      "loss": 0.005,
+      "step": 9996
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001958470973087412,
+      "loss": 0.0023,
+      "step": 9997
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019582863924910113,
+      "loss": 0.0078,
+      "step": 9998
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019581018042403634,
+      "loss": 0.0008,
+      "step": 9999
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019579172083385507,
+      "loss": 0.0017,
+      "step": 10000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001957732604788656,
+      "loss": 0.0075,
+      "step": 10001
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019575479935937627,
+      "loss": 0.0053,
+      "step": 10002
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019573633747569543,
+      "loss": 0.0061,
+      "step": 10003
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001957178748281314,
+      "loss": 0.0016,
+      "step": 10004
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019569941141699262,
+      "loss": 0.0044,
+      "step": 10005
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019568094724258737,
+      "loss": 0.0056,
+      "step": 10006
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019566248230522406,
+      "loss": 0.0089,
+      "step": 10007
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019564401660521113,
+      "loss": 0.0122,
+      "step": 10008
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019562555014285691,
+      "loss": 0.0118,
+      "step": 10009
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019560708291846993,
+      "loss": 0.0094,
+      "step": 10010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019558861493235854,
+      "loss": 0.0094,
+      "step": 10011
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019557014618483128,
+      "loss": 0.0201,
+      "step": 10012
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019555167667619652,
+      "loss": 0.0176,
+      "step": 10013
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019553320640676276,
+      "loss": 0.0116,
+      "step": 10014
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019551473537683854,
+      "loss": 0.0049,
+      "step": 10015
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019549626358673233,
+      "loss": 0.0021,
+      "step": 10016
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001954777910367526,
+      "loss": 0.0081,
+      "step": 10017
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019545931772720792,
+      "loss": 0.0046,
+      "step": 10018
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019544084365840682,
+      "loss": 0.0228,
+      "step": 10019
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019542236883065788,
+      "loss": 0.0279,
+      "step": 10020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019540389324426962,
+      "loss": 0.0058,
+      "step": 10021
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019538541689955062,
+      "loss": 0.0031,
+      "step": 10022
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019536693979680953,
+      "loss": 0.0085,
+      "step": 10023
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019534846193635486,
+      "loss": 0.0039,
+      "step": 10024
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019532998331849534,
+      "loss": 0.0067,
+      "step": 10025
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001953115039435395,
+      "loss": 0.0239,
+      "step": 10026
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000195293023811796,
+      "loss": 0.0246,
+      "step": 10027
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019527454292357353,
+      "loss": 0.0015,
+      "step": 10028
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001952560612791807,
+      "loss": 0.0019,
+      "step": 10029
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001952375788789262,
+      "loss": 0.0039,
+      "step": 10030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019521909572311878,
+      "loss": 0.0141,
+      "step": 10031
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001952006118120671,
+      "loss": 0.0063,
+      "step": 10032
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019518212714607992,
+      "loss": 0.0087,
+      "step": 10033
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019516364172546586,
+      "loss": 0.0031,
+      "step": 10034
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019514515555053377,
+      "loss": 0.0055,
+      "step": 10035
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019512666862159234,
+      "loss": 0.0199,
+      "step": 10036
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019510818093895042,
+      "loss": 0.0089,
+      "step": 10037
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019508969250291668,
+      "loss": 0.0018,
+      "step": 10038
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019507120331379996,
+      "loss": 0.0048,
+      "step": 10039
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019505271337190905,
+      "loss": 0.0018,
+      "step": 10040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019503422267755282,
+      "loss": 0.004,
+      "step": 10041
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019501573123104006,
+      "loss": 0.0058,
+      "step": 10042
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001949972390326796,
+      "loss": 0.0059,
+      "step": 10043
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019497874608278033,
+      "loss": 0.0204,
+      "step": 10044
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019496025238165108,
+      "loss": 0.0083,
+      "step": 10045
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019494175792960077,
+      "loss": 0.025,
+      "step": 10046
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019492326272693824,
+      "loss": 0.0053,
+      "step": 10047
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019490476677397244,
+      "loss": 0.017,
+      "step": 10048
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019488627007101226,
+      "loss": 0.0316,
+      "step": 10049
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019486777261836668,
+      "loss": 0.0043,
+      "step": 10050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019484927441634455,
+      "loss": 0.006,
+      "step": 10051
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001948307754652549,
+      "loss": 0.0116,
+      "step": 10052
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019481227576540668,
+      "loss": 0.0103,
+      "step": 10053
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019479377531710884,
+      "loss": 0.0027,
+      "step": 10054
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019477527412067042,
+      "loss": 0.0059,
+      "step": 10055
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001947567721764004,
+      "loss": 0.0078,
+      "step": 10056
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001947382694846078,
+      "loss": 0.0104,
+      "step": 10057
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019471976604560167,
+      "loss": 0.0038,
+      "step": 10058
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000194701261859691,
+      "loss": 0.0396,
+      "step": 10059
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001946827569271849,
+      "loss": 0.0086,
+      "step": 10060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001946642512483924,
+      "loss": 0.0021,
+      "step": 10061
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001946457448236226,
+      "loss": 0.0071,
+      "step": 10062
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019462723765318457,
+      "loss": 0.0046,
+      "step": 10063
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019460872973738744,
+      "loss": 0.0066,
+      "step": 10064
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019459022107654037,
+      "loss": 0.0093,
+      "step": 10065
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019457171167095238,
+      "loss": 0.023,
+      "step": 10066
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001945532015209327,
+      "loss": 0.0035,
+      "step": 10067
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019453469062679043,
+      "loss": 0.0048,
+      "step": 10068
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019451617898883476,
+      "loss": 0.0019,
+      "step": 10069
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001944976666073749,
+      "loss": 0.0021,
+      "step": 10070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019447915348271997,
+      "loss": 0.0101,
+      "step": 10071
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019446063961517924,
+      "loss": 0.0047,
+      "step": 10072
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001944421250050619,
+      "loss": 0.0067,
+      "step": 10073
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001944236096526772,
+      "loss": 0.0093,
+      "step": 10074
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019440509355833434,
+      "loss": 0.0089,
+      "step": 10075
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001943865767223426,
+      "loss": 0.0148,
+      "step": 10076
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019436805914501126,
+      "loss": 0.0006,
+      "step": 10077
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019434954082664954,
+      "loss": 0.0019,
+      "step": 10078
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019433102176756684,
+      "loss": 0.0169,
+      "step": 10079
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019431250196807237,
+      "loss": 0.0146,
+      "step": 10080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019429398142847544,
+      "loss": 0.0259,
+      "step": 10081
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001942754601490854,
+      "loss": 0.0049,
+      "step": 10082
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019425693813021166,
+      "loss": 0.0134,
+      "step": 10083
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019423841537216348,
+      "loss": 0.0109,
+      "step": 10084
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019421989187525025,
+      "loss": 0.0206,
+      "step": 10085
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019420136763978138,
+      "loss": 0.0064,
+      "step": 10086
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019418284266606618,
+      "loss": 0.0054,
+      "step": 10087
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001941643169544142,
+      "loss": 0.0188,
+      "step": 10088
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001941457905051347,
+      "loss": 0.0113,
+      "step": 10089
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019412726331853716,
+      "loss": 0.0104,
+      "step": 10090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019410873539493103,
+      "loss": 0.0156,
+      "step": 10091
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019409020673462576,
+      "loss": 0.0081,
+      "step": 10092
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001940716773379308,
+      "loss": 0.004,
+      "step": 10093
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001940531472051557,
+      "loss": 0.0102,
+      "step": 10094
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001940346163366098,
+      "loss": 0.0018,
+      "step": 10095
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019401608473260275,
+      "loss": 0.001,
+      "step": 10096
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000193997552393444,
+      "loss": 0.0239,
+      "step": 10097
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019397901931944304,
+      "loss": 0.0114,
+      "step": 10098
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019396048551090944,
+      "loss": 0.0162,
+      "step": 10099
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019394195096815277,
+      "loss": 0.0025,
+      "step": 10100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019392341569148252,
+      "loss": 0.019,
+      "step": 10101
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001939048796812084,
+      "loss": 0.0128,
+      "step": 10102
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019388634293763987,
+      "loss": 0.0122,
+      "step": 10103
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019386780546108655,
+      "loss": 0.0181,
+      "step": 10104
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019384926725185807,
+      "loss": 0.0034,
+      "step": 10105
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019383072831026413,
+      "loss": 0.0096,
+      "step": 10106
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019381218863661422,
+      "loss": 0.0096,
+      "step": 10107
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019379364823121813,
+      "loss": 0.0087,
+      "step": 10108
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001937751070943854,
+      "loss": 0.0067,
+      "step": 10109
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019375656522642573,
+      "loss": 0.0068,
+      "step": 10110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019373802262764888,
+      "loss": 0.0178,
+      "step": 10111
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019371947929836442,
+      "loss": 0.0086,
+      "step": 10112
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019370093523888213,
+      "loss": 0.0101,
+      "step": 10113
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019368239044951176,
+      "loss": 0.0099,
+      "step": 10114
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019366384493056305,
+      "loss": 0.0165,
+      "step": 10115
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019364529868234567,
+      "loss": 0.0138,
+      "step": 10116
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019362675170516943,
+      "loss": 0.0116,
+      "step": 10117
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001936082039993441,
+      "loss": 0.01,
+      "step": 10118
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019358965556517938,
+      "loss": 0.0075,
+      "step": 10119
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001935711064029852,
+      "loss": 0.0129,
+      "step": 10120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001935525565130713,
+      "loss": 0.0076,
+      "step": 10121
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019353400589574744,
+      "loss": 0.0093,
+      "step": 10122
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001935154545513235,
+      "loss": 0.0049,
+      "step": 10123
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019349690248010934,
+      "loss": 0.0012,
+      "step": 10124
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019347834968241488,
+      "loss": 0.0016,
+      "step": 10125
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019345979615854983,
+      "loss": 0.0102,
+      "step": 10126
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019344124190882417,
+      "loss": 0.0007,
+      "step": 10127
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019342268693354776,
+      "loss": 0.0076,
+      "step": 10128
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019340413123303054,
+      "loss": 0.0103,
+      "step": 10129
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001933855748075824,
+      "loss": 0.0014,
+      "step": 10130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019336701765751324,
+      "loss": 0.0053,
+      "step": 10131
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019334845978313301,
+      "loss": 0.0093,
+      "step": 10132
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001933299011847517,
+      "loss": 0.0112,
+      "step": 10133
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019331134186267922,
+      "loss": 0.0046,
+      "step": 10134
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001932927818172256,
+      "loss": 0.02,
+      "step": 10135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019327422104870078,
+      "loss": 0.029,
+      "step": 10136
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001932556595574148,
+      "loss": 0.0202,
+      "step": 10137
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019323709734367765,
+      "loss": 0.0102,
+      "step": 10138
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019321853440779942,
+      "loss": 0.012,
+      "step": 10139
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019319997075009,
+      "loss": 0.0095,
+      "step": 10140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019318140637085955,
+      "loss": 0.0018,
+      "step": 10141
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001931628412704181,
+      "loss": 0.0065,
+      "step": 10142
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019314427544907573,
+      "loss": 0.0099,
+      "step": 10143
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001931257089071425,
+      "loss": 0.0054,
+      "step": 10144
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019310714164492857,
+      "loss": 0.0205,
+      "step": 10145
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019308857366274397,
+      "loss": 0.0112,
+      "step": 10146
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019307000496089887,
+      "loss": 0.0077,
+      "step": 10147
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001930514355397034,
+      "loss": 0.0076,
+      "step": 10148
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019303286539946768,
+      "loss": 0.013,
+      "step": 10149
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019301429454050186,
+      "loss": 0.0143,
+      "step": 10150
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001929957229631161,
+      "loss": 0.0094,
+      "step": 10151
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019297715066762068,
+      "loss": 0.0037,
+      "step": 10152
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019295857765432565,
+      "loss": 0.007,
+      "step": 10153
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019294000392354132,
+      "loss": 0.0053,
+      "step": 10154
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019292142947557783,
+      "loss": 0.0096,
+      "step": 10155
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019290285431074547,
+      "loss": 0.0117,
+      "step": 10156
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019288427842935451,
+      "loss": 0.0118,
+      "step": 10157
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928657018317151,
+      "loss": 0.0038,
+      "step": 10158
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928471245181376,
+      "loss": 0.0266,
+      "step": 10159
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019282854648893216,
+      "loss": 0.0088,
+      "step": 10160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928099677444092,
+      "loss": 0.006,
+      "step": 10161
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019279138828487899,
+      "loss": 0.0027,
+      "step": 10162
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019277280811065182,
+      "loss": 0.0051,
+      "step": 10163
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019275422722203796,
+      "loss": 0.0027,
+      "step": 10164
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019273564561934778,
+      "loss": 0.0023,
+      "step": 10165
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019271706330289173,
+      "loss": 0.0137,
+      "step": 10166
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019269848027298004,
+      "loss": 0.0108,
+      "step": 10167
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019267989652992318,
+      "loss": 0.0063,
+      "step": 10168
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019266131207403144,
+      "loss": 0.0086,
+      "step": 10169
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019264272690561527,
+      "loss": 0.0039,
+      "step": 10170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019262414102498506,
+      "loss": 0.0089,
+      "step": 10171
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019260555443245127,
+      "loss": 0.0089,
+      "step": 10172
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019258696712832425,
+      "loss": 0.003,
+      "step": 10173
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019256837911291454,
+      "loss": 0.0064,
+      "step": 10174
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001925497903865325,
+      "loss": 0.0039,
+      "step": 10175
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019253120094948865,
+      "loss": 0.0042,
+      "step": 10176
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001925126108020935,
+      "loss": 0.0585,
+      "step": 10177
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019249401994465745,
+      "loss": 0.0068,
+      "step": 10178
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019247542837749103,
+      "loss": 0.0037,
+      "step": 10179
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019245683610090485,
+      "loss": 0.0227,
+      "step": 10180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019243824311520935,
+      "loss": 0.0078,
+      "step": 10181
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019241964942071508,
+      "loss": 0.0047,
+      "step": 10182
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001924010550177326,
+      "loss": 0.0092,
+      "step": 10183
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001923824599065724,
+      "loss": 0.0056,
+      "step": 10184
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019236386408754517,
+      "loss": 0.0019,
+      "step": 10185
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019234526756096144,
+      "loss": 0.0318,
+      "step": 10186
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001923266703271318,
+      "loss": 0.0074,
+      "step": 10187
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019230807238636687,
+      "loss": 0.0029,
+      "step": 10188
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019228947373897727,
+      "loss": 0.0128,
+      "step": 10189
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019227087438527363,
+      "loss": 0.0031,
+      "step": 10190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001922522743255666,
+      "loss": 0.0128,
+      "step": 10191
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019223367356016686,
+      "loss": 0.0038,
+      "step": 10192
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000192215072089385,
+      "loss": 0.0172,
+      "step": 10193
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019219646991353178,
+      "loss": 0.0099,
+      "step": 10194
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019217786703291785,
+      "loss": 0.0029,
+      "step": 10195
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001921592634478539,
+      "loss": 0.008,
+      "step": 10196
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001921406591586507,
+      "loss": 0.0151,
+      "step": 10197
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019212205416561893,
+      "loss": 0.0105,
+      "step": 10198
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001921034484690694,
+      "loss": 0.0055,
+      "step": 10199
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019208484206931272,
+      "loss": 0.0143,
+      "step": 10200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019206623496665978,
+      "loss": 0.0009,
+      "step": 10201
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019204762716142133,
+      "loss": 0.0083,
+      "step": 10202
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001920290186539081,
+      "loss": 0.0032,
+      "step": 10203
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019201040944443094,
+      "loss": 0.0144,
+      "step": 10204
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019199179953330065,
+      "loss": 0.0028,
+      "step": 10205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000191973188920828,
+      "loss": 0.0013,
+      "step": 10206
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019195457760732388,
+      "loss": 0.0021,
+      "step": 10207
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019193596559309914,
+      "loss": 0.0149,
+      "step": 10208
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019191735287846462,
+      "loss": 0.023,
+      "step": 10209
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019189873946373116,
+      "loss": 0.0063,
+      "step": 10210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019188012534920968,
+      "loss": 0.0163,
+      "step": 10211
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019186151053521104,
+      "loss": 0.0012,
+      "step": 10212
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001918428950220462,
+      "loss": 0.0127,
+      "step": 10213
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019182427881002595,
+      "loss": 0.0096,
+      "step": 10214
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019180566189946134,
+      "loss": 0.0136,
+      "step": 10215
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019178704429066327,
+      "loss": 0.0088,
+      "step": 10216
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019176842598394266,
+      "loss": 0.0047,
+      "step": 10217
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001917498069796105,
+      "loss": 0.0073,
+      "step": 10218
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019173118727797774,
+      "loss": 0.0011,
+      "step": 10219
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019171256687935538,
+      "loss": 0.0053,
+      "step": 10220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019169394578405445,
+      "loss": 0.0057,
+      "step": 10221
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916753239923859,
+      "loss": 0.0093,
+      "step": 10222
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916567015046608,
+      "loss": 0.0069,
+      "step": 10223
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916380783211901,
+      "loss": 0.0355,
+      "step": 10224
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019161945444228493,
+      "loss": 0.0062,
+      "step": 10225
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916008298682563,
+      "loss": 0.0207,
+      "step": 10226
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001915822045994153,
+      "loss": 0.0122,
+      "step": 10227
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019156357863607296,
+      "loss": 0.0064,
+      "step": 10228
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019154495197854038,
+      "loss": 0.0067,
+      "step": 10229
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019152632462712872,
+      "loss": 0.0032,
+      "step": 10230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019150769658214908,
+      "loss": 0.0061,
+      "step": 10231
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019148906784391252,
+      "loss": 0.0076,
+      "step": 10232
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001914704384127302,
+      "loss": 0.0018,
+      "step": 10233
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001914518082889133,
+      "loss": 0.0192,
+      "step": 10234
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019143317747277294,
+      "loss": 0.0071,
+      "step": 10235
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019141454596462035,
+      "loss": 0.0047,
+      "step": 10236
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019139591376476656,
+      "loss": 0.0155,
+      "step": 10237
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000191377280873523,
+      "loss": 0.0049,
+      "step": 10238
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019135864729120065,
+      "loss": 0.0078,
+      "step": 10239
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019134001301811086,
+      "loss": 0.0117,
+      "step": 10240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019132137805456483,
+      "loss": 0.0022,
+      "step": 10241
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019130274240087376,
+      "loss": 0.0068,
+      "step": 10242
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019128410605734892,
+      "loss": 0.0041,
+      "step": 10243
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001912654690243016,
+      "loss": 0.0152,
+      "step": 10244
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019124683130204308,
+      "loss": 0.0063,
+      "step": 10245
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019122819289088455,
+      "loss": 0.0204,
+      "step": 10246
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019120955379113745,
+      "loss": 0.002,
+      "step": 10247
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001911909140031129,
+      "loss": 0.0244,
+      "step": 10248
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001911722735271224,
+      "loss": 0.0077,
+      "step": 10249
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001911536323634773,
+      "loss": 0.0069,
+      "step": 10250
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019113499051248877,
+      "loss": 0.0085,
+      "step": 10251
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019111634797446824,
+      "loss": 0.0161,
+      "step": 10252
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001910977047497271,
+      "loss": 0.0056,
+      "step": 10253
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019107906083857672,
+      "loss": 0.018,
+      "step": 10254
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001910604162413285,
+      "loss": 0.0025,
+      "step": 10255
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019104177095829382,
+      "loss": 0.0123,
+      "step": 10256
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019102312498978408,
+      "loss": 0.0046,
+      "step": 10257
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019100447833611067,
+      "loss": 0.0063,
+      "step": 10258
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909858309975851,
+      "loss": 0.0041,
+      "step": 10259
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909671829745188,
+      "loss": 0.0083,
+      "step": 10260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909485342672232,
+      "loss": 0.0067,
+      "step": 10261
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909298848760098,
+      "loss": 0.0089,
+      "step": 10262
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019091123480119004,
+      "loss": 0.0136,
+      "step": 10263
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001908925840430755,
+      "loss": 0.0045,
+      "step": 10264
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019087393260197754,
+      "loss": 0.0056,
+      "step": 10265
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019085528047820777,
+      "loss": 0.0043,
+      "step": 10266
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001908366276720777,
+      "loss": 0.0166,
+      "step": 10267
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001908179741838989,
+      "loss": 0.0056,
+      "step": 10268
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019079932001398283,
+      "loss": 0.0074,
+      "step": 10269
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001907806651626411,
+      "loss": 0.0041,
+      "step": 10270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001907620096301853,
+      "loss": 0.0244,
+      "step": 10271
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019074335341692696,
+      "loss": 0.0375,
+      "step": 10272
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019072469652317778,
+      "loss": 0.0141,
+      "step": 10273
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001907060389492493,
+      "loss": 0.0035,
+      "step": 10274
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019068738069545307,
+      "loss": 0.0059,
+      "step": 10275
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019066872176210078,
+      "loss": 0.0107,
+      "step": 10276
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001906500621495041,
+      "loss": 0.0074,
+      "step": 10277
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019063140185797463,
+      "loss": 0.0065,
+      "step": 10278
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019061274088782407,
+      "loss": 0.0134,
+      "step": 10279
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019059407923936408,
+      "loss": 0.0018,
+      "step": 10280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001905754169129063,
+      "loss": 0.0137,
+      "step": 10281
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001905567539087625,
+      "loss": 0.0018,
+      "step": 10282
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019053809022724437,
+      "loss": 0.0075,
+      "step": 10283
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019051942586866357,
+      "loss": 0.0085,
+      "step": 10284
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019050076083333188,
+      "loss": 0.0143,
+      "step": 10285
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019048209512156106,
+      "loss": 0.0224,
+      "step": 10286
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019046342873366284,
+      "loss": 0.0119,
+      "step": 10287
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019044476166994896,
+      "loss": 0.0061,
+      "step": 10288
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019042609393073122,
+      "loss": 0.0056,
+      "step": 10289
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019040742551632136,
+      "loss": 0.016,
+      "step": 10290
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019038875642703122,
+      "loss": 0.0036,
+      "step": 10291
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001903700866631727,
+      "loss": 0.0243,
+      "step": 10292
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019035141622505747,
+      "loss": 0.0052,
+      "step": 10293
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019033274511299743,
+      "loss": 0.0078,
+      "step": 10294
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001903140733273044,
+      "loss": 0.0066,
+      "step": 10295
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019029540086829027,
+      "loss": 0.0061,
+      "step": 10296
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019027672773626686,
+      "loss": 0.0051,
+      "step": 10297
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019025805393154606,
+      "loss": 0.0049,
+      "step": 10298
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001902393794544398,
+      "loss": 0.0102,
+      "step": 10299
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019022070430525992,
+      "loss": 0.0113,
+      "step": 10300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019020202848431836,
+      "loss": 0.0076,
+      "step": 10301
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019018335199192705,
+      "loss": 0.0111,
+      "step": 10302
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019016467482839788,
+      "loss": 0.0253,
+      "step": 10303
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019014599699404284,
+      "loss": 0.001,
+      "step": 10304
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001901273184891739,
+      "loss": 0.004,
+      "step": 10305
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019010863931410296,
+      "loss": 0.0116,
+      "step": 10306
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019008995946914205,
+      "loss": 0.0141,
+      "step": 10307
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019007127895460317,
+      "loss": 0.0184,
+      "step": 10308
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019005259777079824,
+      "loss": 0.0119,
+      "step": 10309
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019003391591803934,
+      "loss": 0.0037,
+      "step": 10310
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019001523339663844,
+      "loss": 0.0187,
+      "step": 10311
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018999655020690765,
+      "loss": 0.0081,
+      "step": 10312
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018997786634915897,
+      "loss": 0.0083,
+      "step": 10313
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018995918182370446,
+      "loss": 0.0095,
+      "step": 10314
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001899404966308562,
+      "loss": 0.0098,
+      "step": 10315
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001899218107709262,
+      "loss": 0.0076,
+      "step": 10316
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018990312424422664,
+      "loss": 0.0191,
+      "step": 10317
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018988443705106957,
+      "loss": 0.0144,
+      "step": 10318
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001898657491917672,
+      "loss": 0.0093,
+      "step": 10319
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018984706066663143,
+      "loss": 0.0148,
+      "step": 10320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001898283714759746,
+      "loss": 0.0095,
+      "step": 10321
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018980968162010876,
+      "loss": 0.0058,
+      "step": 10322
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018979099109934608,
+      "loss": 0.0272,
+      "step": 10323
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018977229991399883,
+      "loss": 0.0076,
+      "step": 10324
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000189753608064379,
+      "loss": 0.009,
+      "step": 10325
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001897349155507989,
+      "loss": 0.0059,
+      "step": 10326
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001897162223735707,
+      "loss": 0.0151,
+      "step": 10327
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018969752853300662,
+      "loss": 0.0113,
+      "step": 10328
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018967883402941891,
+      "loss": 0.0065,
+      "step": 10329
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018966013886311976,
+      "loss": 0.0046,
+      "step": 10330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018964144303442137,
+      "loss": 0.0102,
+      "step": 10331
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001896227465436361,
+      "loss": 0.0056,
+      "step": 10332
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001896040493910762,
+      "loss": 0.0043,
+      "step": 10333
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018958535157705386,
+      "loss": 0.022,
+      "step": 10334
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018956665310188145,
+      "loss": 0.0155,
+      "step": 10335
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018954795396587123,
+      "loss": 0.0062,
+      "step": 10336
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018952925416933556,
+      "loss": 0.0031,
+      "step": 10337
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018951055371258673,
+      "loss": 0.0344,
+      "step": 10338
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018949185259593707,
+      "loss": 0.0102,
+      "step": 10339
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018947315081969889,
+      "loss": 0.0071,
+      "step": 10340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018945444838418458,
+      "loss": 0.0069,
+      "step": 10341
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018943574528970655,
+      "loss": 0.0144,
+      "step": 10342
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001894170415365771,
+      "loss": 0.006,
+      "step": 10343
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018939833712510866,
+      "loss": 0.0074,
+      "step": 10344
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018937963205561362,
+      "loss": 0.0066,
+      "step": 10345
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018936092632840435,
+      "loss": 0.0097,
+      "step": 10346
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001893422199437934,
+      "loss": 0.006,
+      "step": 10347
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018932351290209306,
+      "loss": 0.0052,
+      "step": 10348
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018930480520361584,
+      "loss": 0.0054,
+      "step": 10349
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018928609684867417,
+      "loss": 0.0093,
+      "step": 10350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001892673878375805,
+      "loss": 0.0018,
+      "step": 10351
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018924867817064734,
+      "loss": 0.0034,
+      "step": 10352
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018922996784818721,
+      "loss": 0.0085,
+      "step": 10353
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018921125687051252,
+      "loss": 0.0112,
+      "step": 10354
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018919254523793586,
+      "loss": 0.0079,
+      "step": 10355
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018917383295076972,
+      "loss": 0.0176,
+      "step": 10356
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018915512000932657,
+      "loss": 0.02,
+      "step": 10357
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018913640641391905,
+      "loss": 0.0117,
+      "step": 10358
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018911769216485965,
+      "loss": 0.0123,
+      "step": 10359
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018909897726246096,
+      "loss": 0.0069,
+      "step": 10360
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018908026170703552,
+      "loss": 0.0091,
+      "step": 10361
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018906154549889597,
+      "loss": 0.0069,
+      "step": 10362
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018904282863835487,
+      "loss": 0.0125,
+      "step": 10363
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001890241111257248,
+      "loss": 0.0521,
+      "step": 10364
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018900539296131847,
+      "loss": 0.0075,
+      "step": 10365
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018898667414544843,
+      "loss": 0.0062,
+      "step": 10366
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001889679546784273,
+      "loss": 0.0049,
+      "step": 10367
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018894923456056782,
+      "loss": 0.0096,
+      "step": 10368
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018893051379218257,
+      "loss": 0.0149,
+      "step": 10369
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001889117923735843,
+      "loss": 0.0024,
+      "step": 10370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001888930703050856,
+      "loss": 0.0177,
+      "step": 10371
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001888743475869992,
+      "loss": 0.0161,
+      "step": 10372
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018885562421963782,
+      "loss": 0.0013,
+      "step": 10373
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018883690020331416,
+      "loss": 0.0155,
+      "step": 10374
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018881817553834102,
+      "loss": 0.0089,
+      "step": 10375
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018879945022503102,
+      "loss": 0.0067,
+      "step": 10376
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000188780724263697,
+      "loss": 0.0135,
+      "step": 10377
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018876199765465164,
+      "loss": 0.007,
+      "step": 10378
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001887432703982078,
+      "loss": 0.0049,
+      "step": 10379
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001887245424946782,
+      "loss": 0.0039,
+      "step": 10380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018870581394437558,
+      "loss": 0.0154,
+      "step": 10381
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018868708474761286,
+      "loss": 0.0085,
+      "step": 10382
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018866835490470276,
+      "loss": 0.0042,
+      "step": 10383
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001886496244159582,
+      "loss": 0.0111,
+      "step": 10384
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001886308932816919,
+      "loss": 0.0057,
+      "step": 10385
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018861216150221677,
+      "loss": 0.0122,
+      "step": 10386
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018859342907784566,
+      "loss": 0.0071,
+      "step": 10387
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018857469600889145,
+      "loss": 0.0243,
+      "step": 10388
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018855596229566706,
+      "loss": 0.0099,
+      "step": 10389
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018853722793848524,
+      "loss": 0.007,
+      "step": 10390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018851849293765894,
+      "loss": 0.0115,
+      "step": 10391
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018849975729350117,
+      "loss": 0.0111,
+      "step": 10392
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018848102100632477,
+      "loss": 0.0239,
+      "step": 10393
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018846228407644265,
+      "loss": 0.0033,
+      "step": 10394
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001884435465041678,
+      "loss": 0.0059,
+      "step": 10395
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018842480828981312,
+      "loss": 0.0012,
+      "step": 10396
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018840606943369166,
+      "loss": 0.0064,
+      "step": 10397
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018838732993611636,
+      "loss": 0.0055,
+      "step": 10398
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018836858979740016,
+      "loss": 0.0129,
+      "step": 10399
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018834984901785608,
+      "loss": 0.0296,
+      "step": 10400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001883311075977971,
+      "loss": 0.0181,
+      "step": 10401
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018831236553753632,
+      "loss": 0.0011,
+      "step": 10402
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001882936228373867,
+      "loss": 0.0038,
+      "step": 10403
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001882748794976613,
+      "loss": 0.0334,
+      "step": 10404
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018825613551867311,
+      "loss": 0.0026,
+      "step": 10405
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001882373909007353,
+      "loss": 0.0058,
+      "step": 10406
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001882186456441609,
+      "loss": 0.0099,
+      "step": 10407
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018819989974926292,
+      "loss": 0.0099,
+      "step": 10408
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018818115321635455,
+      "loss": 0.0036,
+      "step": 10409
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018816240604574883,
+      "loss": 0.0056,
+      "step": 10410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018814365823775889,
+      "loss": 0.0139,
+      "step": 10411
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018812490979269787,
+      "loss": 0.0089,
+      "step": 10412
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001881061607108789,
+      "loss": 0.0123,
+      "step": 10413
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018808741099261508,
+      "loss": 0.0127,
+      "step": 10414
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018806866063821961,
+      "loss": 0.0235,
+      "step": 10415
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018804990964800566,
+      "loss": 0.0093,
+      "step": 10416
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018803115802228638,
+      "loss": 0.0086,
+      "step": 10417
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000188012405761375,
+      "loss": 0.0047,
+      "step": 10418
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001879936528655847,
+      "loss": 0.0054,
+      "step": 10419
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018797489933522862,
+      "loss": 0.0042,
+      "step": 10420
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001879561451706201,
+      "loss": 0.0228,
+      "step": 10421
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001879373903720723,
+      "loss": 0.0081,
+      "step": 10422
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018791863493989843,
+      "loss": 0.0169,
+      "step": 10423
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018789987887441178,
+      "loss": 0.0091,
+      "step": 10424
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001878811221759256,
+      "loss": 0.0057,
+      "step": 10425
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018786236484475324,
+      "loss": 0.0058,
+      "step": 10426
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018784360688120784,
+      "loss": 0.0046,
+      "step": 10427
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018782484828560282,
+      "loss": 0.0087,
+      "step": 10428
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018780608905825144,
+      "loss": 0.0032,
+      "step": 10429
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018778732919946703,
+      "loss": 0.0127,
+      "step": 10430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018776856870956286,
+      "loss": 0.0372,
+      "step": 10431
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018774980758885229,
+      "loss": 0.0059,
+      "step": 10432
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018773104583764865,
+      "loss": 0.011,
+      "step": 10433
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018771228345626536,
+      "loss": 0.0169,
+      "step": 10434
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018769352044501573,
+      "loss": 0.0068,
+      "step": 10435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001876747568042132,
+      "loss": 0.0115,
+      "step": 10436
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018765599253417104,
+      "loss": 0.0044,
+      "step": 10437
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018763722763520278,
+      "loss": 0.004,
+      "step": 10438
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018761846210762175,
+      "loss": 0.004,
+      "step": 10439
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018759969595174146,
+      "loss": 0.0028,
+      "step": 10440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018758092916787523,
+      "loss": 0.0063,
+      "step": 10441
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018756216175633652,
+      "loss": 0.0293,
+      "step": 10442
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018754339371743878,
+      "loss": 0.0059,
+      "step": 10443
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018752462505149556,
+      "loss": 0.0042,
+      "step": 10444
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018750585575882023,
+      "loss": 0.0066,
+      "step": 10445
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018748708583972632,
+      "loss": 0.0147,
+      "step": 10446
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018746831529452732,
+      "loss": 0.0048,
+      "step": 10447
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018744954412353671,
+      "loss": 0.0083,
+      "step": 10448
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001874307723270681,
+      "loss": 0.0006,
+      "step": 10449
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018741199990543486,
+      "loss": 0.0197,
+      "step": 10450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001873932268589506,
+      "loss": 0.0052,
+      "step": 10451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001873744531879289,
+      "loss": 0.0087,
+      "step": 10452
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018735567889268322,
+      "loss": 0.0167,
+      "step": 10453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001873369039735273,
+      "loss": 0.0044,
+      "step": 10454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018731812843077454,
+      "loss": 0.0184,
+      "step": 10455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018729935226473855,
+      "loss": 0.0094,
+      "step": 10456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018728057547573302,
+      "loss": 0.0078,
+      "step": 10457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018726179806407151,
+      "loss": 0.0024,
+      "step": 10458
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018724302003006763,
+      "loss": 0.0192,
+      "step": 10459
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000187224241374035,
+      "loss": 0.002,
+      "step": 10460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018720546209628728,
+      "loss": 0.0061,
+      "step": 10461
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001871866821971381,
+      "loss": 0.0185,
+      "step": 10462
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001871679016769012,
+      "loss": 0.0105,
+      "step": 10463
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018714912053589011,
+      "loss": 0.0065,
+      "step": 10464
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001871303387744186,
+      "loss": 0.012,
+      "step": 10465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018711155639280035,
+      "loss": 0.004,
+      "step": 10466
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018709277339134907,
+      "loss": 0.0092,
+      "step": 10467
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018707398977037845,
+      "loss": 0.0121,
+      "step": 10468
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001870552055302022,
+      "loss": 0.0158,
+      "step": 10469
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018703642067113407,
+      "loss": 0.0084,
+      "step": 10470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001870176351934878,
+      "loss": 0.0224,
+      "step": 10471
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018699884909757718,
+      "loss": 0.0139,
+      "step": 10472
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018698006238371593,
+      "loss": 0.014,
+      "step": 10473
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018696127505221782,
+      "loss": 0.0163,
+      "step": 10474
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018694248710339665,
+      "loss": 0.0077,
+      "step": 10475
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001869236985375662,
+      "loss": 0.0111,
+      "step": 10476
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018690490935504027,
+      "loss": 0.0074,
+      "step": 10477
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018688611955613275,
+      "loss": 0.0088,
+      "step": 10478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018686732914115737,
+      "loss": 0.004,
+      "step": 10479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000186848538110428,
+      "loss": 0.0109,
+      "step": 10480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018682974646425852,
+      "loss": 0.0056,
+      "step": 10481
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001868109542029627,
+      "loss": 0.0022,
+      "step": 10482
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018679216132685451,
+      "loss": 0.0141,
+      "step": 10483
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018677336783624776,
+      "loss": 0.0074,
+      "step": 10484
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001867545737314563,
+      "loss": 0.005,
+      "step": 10485
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018673577901279415,
+      "loss": 0.0014,
+      "step": 10486
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001867169836805751,
+      "loss": 0.0071,
+      "step": 10487
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018669818773511313,
+      "loss": 0.0195,
+      "step": 10488
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018667939117672215,
+      "loss": 0.0078,
+      "step": 10489
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018666059400571612,
+      "loss": 0.0117,
+      "step": 10490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000186641796222409,
+      "loss": 0.0094,
+      "step": 10491
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018662299782711464,
+      "loss": 0.0057,
+      "step": 10492
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018660419882014713,
+      "loss": 0.0097,
+      "step": 10493
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018658539920182038,
+      "loss": 0.0039,
+      "step": 10494
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018656659897244842,
+      "loss": 0.009,
+      "step": 10495
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018654779813234525,
+      "loss": 0.0098,
+      "step": 10496
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018652899668182481,
+      "loss": 0.0045,
+      "step": 10497
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018651019462120118,
+      "loss": 0.0109,
+      "step": 10498
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001864913919507884,
+      "loss": 0.0123,
+      "step": 10499
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018647258867090054,
+      "loss": 0.0099,
+      "step": 10500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018645378478185156,
+      "loss": 0.0032,
+      "step": 10501
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018643498028395554,
+      "loss": 0.0104,
+      "step": 10502
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001864161751775266,
+      "loss": 0.0022,
+      "step": 10503
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001863973694628788,
+      "loss": 0.0224,
+      "step": 10504
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018637856314032624,
+      "loss": 0.0068,
+      "step": 10505
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018635975621018297,
+      "loss": 0.021,
+      "step": 10506
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018634094867276313,
+      "loss": 0.0235,
+      "step": 10507
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018632214052838087,
+      "loss": 0.0012,
+      "step": 10508
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001863033317773503,
+      "loss": 0.0047,
+      "step": 10509
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018628452241998555,
+      "loss": 0.0034,
+      "step": 10510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001862657124566008,
+      "loss": 0.0037,
+      "step": 10511
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018624690188751017,
+      "loss": 0.0203,
+      "step": 10512
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018622809071302786,
+      "loss": 0.0042,
+      "step": 10513
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001862092789334681,
+      "loss": 0.0165,
+      "step": 10514
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018619046654914497,
+      "loss": 0.003,
+      "step": 10515
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001861716535603728,
+      "loss": 0.0154,
+      "step": 10516
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018615283996746565,
+      "loss": 0.0024,
+      "step": 10517
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001861340257707379,
+      "loss": 0.0076,
+      "step": 10518
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001861152109705037,
+      "loss": 0.0014,
+      "step": 10519
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018609639556707726,
+      "loss": 0.0085,
+      "step": 10520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018607757956077292,
+      "loss": 0.0137,
+      "step": 10521
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018605876295190487,
+      "loss": 0.005,
+      "step": 10522
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018603994574078748,
+      "loss": 0.0004,
+      "step": 10523
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018602112792773492,
+      "loss": 0.0051,
+      "step": 10524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018600230951306156,
+      "loss": 0.0127,
+      "step": 10525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001859834904970816,
+      "loss": 0.0021,
+      "step": 10526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018596467088010944,
+      "loss": 0.0077,
+      "step": 10527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018594585066245944,
+      "loss": 0.006,
+      "step": 10528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018592702984444585,
+      "loss": 0.0083,
+      "step": 10529
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018590820842638304,
+      "loss": 0.0064,
+      "step": 10530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001858893864085854,
+      "loss": 0.0051,
+      "step": 10531
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018587056379136725,
+      "loss": 0.0027,
+      "step": 10532
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018585174057504297,
+      "loss": 0.005,
+      "step": 10533
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018583291675992693,
+      "loss": 0.0098,
+      "step": 10534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018581409234633354,
+      "loss": 0.0163,
+      "step": 10535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001857952673345772,
+      "loss": 0.0221,
+      "step": 10536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018577644172497237,
+      "loss": 0.0115,
+      "step": 10537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001857576155178334,
+      "loss": 0.0016,
+      "step": 10538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018573878871347473,
+      "loss": 0.0045,
+      "step": 10539
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018571996131221084,
+      "loss": 0.0058,
+      "step": 10540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018570113331435622,
+      "loss": 0.01,
+      "step": 10541
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018568230472022524,
+      "loss": 0.0113,
+      "step": 10542
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018566347553013242,
+      "loss": 0.0186,
+      "step": 10543
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018564464574439223,
+      "loss": 0.0082,
+      "step": 10544
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018562581536331918,
+      "loss": 0.0038,
+      "step": 10545
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001856069843872278,
+      "loss": 0.0113,
+      "step": 10546
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018558815281643254,
+      "loss": 0.004,
+      "step": 10547
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018556932065124794,
+      "loss": 0.0036,
+      "step": 10548
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018555048789198854,
+      "loss": 0.0018,
+      "step": 10549
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001855316545389689,
+      "loss": 0.012,
+      "step": 10550
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018551282059250354,
+      "loss": 0.0176,
+      "step": 10551
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018549398605290706,
+      "loss": 0.0126,
+      "step": 10552
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018547515092049398,
+      "loss": 0.0115,
+      "step": 10553
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018545631519557894,
+      "loss": 0.0191,
+      "step": 10554
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018543747887847653,
+      "loss": 0.0019,
+      "step": 10555
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018541864196950131,
+      "loss": 0.0044,
+      "step": 10556
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001853998044689679,
+      "loss": 0.0035,
+      "step": 10557
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018538096637719096,
+      "loss": 0.0046,
+      "step": 10558
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001853621276944851,
+      "loss": 0.0164,
+      "step": 10559
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018534328842116494,
+      "loss": 0.0022,
+      "step": 10560
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018532444855754511,
+      "loss": 0.0157,
+      "step": 10561
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018530560810394036,
+      "loss": 0.0078,
+      "step": 10562
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018528676706066528,
+      "loss": 0.0084,
+      "step": 10563
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001852679254280346,
+      "loss": 0.0204,
+      "step": 10564
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018524908320636304,
+      "loss": 0.005,
+      "step": 10565
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018523024039596517,
+      "loss": 0.0035,
+      "step": 10566
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018521139699715584,
+      "loss": 0.0143,
+      "step": 10567
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018519255301024967,
+      "loss": 0.0285,
+      "step": 10568
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018517370843556145,
+      "loss": 0.0018,
+      "step": 10569
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001851548632734059,
+      "loss": 0.0031,
+      "step": 10570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018513601752409778,
+      "loss": 0.0149,
+      "step": 10571
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018511717118795183,
+      "loss": 0.0155,
+      "step": 10572
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018509832426528286,
+      "loss": 0.021,
+      "step": 10573
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018507947675640564,
+      "loss": 0.0194,
+      "step": 10574
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850606286616349,
+      "loss": 0.0064,
+      "step": 10575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850417799812855,
+      "loss": 0.0162,
+      "step": 10576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850229307156722,
+      "loss": 0.0064,
+      "step": 10577
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850040808651099,
+      "loss": 0.0087,
+      "step": 10578
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018498523042991336,
+      "loss": 0.0247,
+      "step": 10579
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001849663794103974,
+      "loss": 0.0237,
+      "step": 10580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018494752780687688,
+      "loss": 0.0049,
+      "step": 10581
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001849286756196667,
+      "loss": 0.0158,
+      "step": 10582
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018490982284908173,
+      "loss": 0.0012,
+      "step": 10583
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001848909694954368,
+      "loss": 0.0019,
+      "step": 10584
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018487211555904686,
+      "loss": 0.0091,
+      "step": 10585
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001848532610402267,
+      "loss": 0.026,
+      "step": 10586
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001848344059392913,
+      "loss": 0.0024,
+      "step": 10587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018481555025655564,
+      "loss": 0.0107,
+      "step": 10588
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018479669399233453,
+      "loss": 0.0263,
+      "step": 10589
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001847778371469429,
+      "loss": 0.0067,
+      "step": 10590
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018475897972069577,
+      "loss": 0.0109,
+      "step": 10591
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018474012171390807,
+      "loss": 0.0435,
+      "step": 10592
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018472126312689478,
+      "loss": 0.042,
+      "step": 10593
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018470240395997086,
+      "loss": 0.0045,
+      "step": 10594
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001846835442134512,
+      "loss": 0.0106,
+      "step": 10595
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018466468388765096,
+      "loss": 0.0103,
+      "step": 10596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001846458229828851,
+      "loss": 0.0079,
+      "step": 10597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018462696149946855,
+      "loss": 0.0029,
+      "step": 10598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001846080994377164,
+      "loss": 0.0062,
+      "step": 10599
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018458923679794362,
+      "loss": 0.0062,
+      "step": 10600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001845703735804653,
+      "loss": 0.0065,
+      "step": 10601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018455150978559652,
+      "loss": 0.0074,
+      "step": 10602
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001845326454136523,
+      "loss": 0.007,
+      "step": 10603
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018451378046494766,
+      "loss": 0.0158,
+      "step": 10604
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001844949149397978,
+      "loss": 0.0039,
+      "step": 10605
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018447604883851774,
+      "loss": 0.0136,
+      "step": 10606
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018445718216142257,
+      "loss": 0.0037,
+      "step": 10607
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018443831490882743,
+      "loss": 0.0047,
+      "step": 10608
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001844194470810474,
+      "loss": 0.0103,
+      "step": 10609
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018440057867839766,
+      "loss": 0.0107,
+      "step": 10610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018438170970119328,
+      "loss": 0.0029,
+      "step": 10611
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018436284014974947,
+      "loss": 0.0113,
+      "step": 10612
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018434397002438138,
+      "loss": 0.0136,
+      "step": 10613
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018432509932540412,
+      "loss": 0.0072,
+      "step": 10614
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018430622805313298,
+      "loss": 0.006,
+      "step": 10615
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018428735620788303,
+      "loss": 0.006,
+      "step": 10616
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001842684837899695,
+      "loss": 0.0062,
+      "step": 10617
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018424961079970762,
+      "loss": 0.0221,
+      "step": 10618
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001842307372374126,
+      "loss": 0.0133,
+      "step": 10619
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018421186310339967,
+      "loss": 0.0015,
+      "step": 10620
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018419298839798402,
+      "loss": 0.0083,
+      "step": 10621
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018417411312148085,
+      "loss": 0.0084,
+      "step": 10622
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018415523727420557,
+      "loss": 0.0089,
+      "step": 10623
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018413636085647334,
+      "loss": 0.009,
+      "step": 10624
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018411748386859946,
+      "loss": 0.0056,
+      "step": 10625
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018409860631089916,
+      "loss": 0.0058,
+      "step": 10626
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018407972818368777,
+      "loss": 0.002,
+      "step": 10627
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018406084948728062,
+      "loss": 0.012,
+      "step": 10628
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000184041970221993,
+      "loss": 0.0205,
+      "step": 10629
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001840230903881402,
+      "loss": 0.0045,
+      "step": 10630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018400420998603755,
+      "loss": 0.0059,
+      "step": 10631
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001839853290160004,
+      "loss": 0.0181,
+      "step": 10632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018396644747834413,
+      "loss": 0.0183,
+      "step": 10633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018394756537338407,
+      "loss": 0.0096,
+      "step": 10634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018392868270143554,
+      "loss": 0.0037,
+      "step": 10635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018390979946281401,
+      "loss": 0.0067,
+      "step": 10636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001838909156578348,
+      "loss": 0.0116,
+      "step": 10637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018387203128681333,
+      "loss": 0.0075,
+      "step": 10638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018385314635006509,
+      "loss": 0.01,
+      "step": 10639
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001838342608479053,
+      "loss": 0.0048,
+      "step": 10640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001838153747806495,
+      "loss": 0.0134,
+      "step": 10641
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001837964881486131,
+      "loss": 0.0074,
+      "step": 10642
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018377760095211157,
+      "loss": 0.0045,
+      "step": 10643
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018375871319146033,
+      "loss": 0.0136,
+      "step": 10644
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001837398248669749,
+      "loss": 0.0082,
+      "step": 10645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018372093597897066,
+      "loss": 0.0064,
+      "step": 10646
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018370204652776317,
+      "loss": 0.0127,
+      "step": 10647
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018368315651366788,
+      "loss": 0.0051,
+      "step": 10648
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018366426593700028,
+      "loss": 0.0013,
+      "step": 10649
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001836453747980759,
+      "loss": 0.0123,
+      "step": 10650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018362648309721023,
+      "loss": 0.0177,
+      "step": 10651
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018360759083471884,
+      "loss": 0.009,
+      "step": 10652
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018358869801091723,
+      "loss": 0.0039,
+      "step": 10653
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018356980462612098,
+      "loss": 0.0083,
+      "step": 10654
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018355091068064557,
+      "loss": 0.0139,
+      "step": 10655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001835320161748067,
+      "loss": 0.0096,
+      "step": 10656
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018351312110891982,
+      "loss": 0.0058,
+      "step": 10657
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018349422548330057,
+      "loss": 0.0093,
+      "step": 10658
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018347532929826455,
+      "loss": 0.0098,
+      "step": 10659
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001834564325541273,
+      "loss": 0.0077,
+      "step": 10660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001834375352512045,
+      "loss": 0.0132,
+      "step": 10661
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018341863738981175,
+      "loss": 0.0029,
+      "step": 10662
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018339973897026464,
+      "loss": 0.0047,
+      "step": 10663
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018338083999287883,
+      "loss": 0.0038,
+      "step": 10664
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018336194045797,
+      "loss": 0.0195,
+      "step": 10665
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018334304036585384,
+      "loss": 0.0097,
+      "step": 10666
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018332413971684593,
+      "loss": 0.0049,
+      "step": 10667
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018330523851126196,
+      "loss": 0.0045,
+      "step": 10668
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018328633674941766,
+      "loss": 0.0032,
+      "step": 10669
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001832674344316287,
+      "loss": 0.0083,
+      "step": 10670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001832485315582108,
+      "loss": 0.0108,
+      "step": 10671
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018322962812947967,
+      "loss": 0.0233,
+      "step": 10672
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018321072414575102,
+      "loss": 0.0047,
+      "step": 10673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018319181960734058,
+      "loss": 0.0049,
+      "step": 10674
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000183172914514564,
+      "loss": 0.006,
+      "step": 10675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018315400886773726,
+      "loss": 0.0111,
+      "step": 10676
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018313510266717596,
+      "loss": 0.007,
+      "step": 10677
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018311619591319592,
+      "loss": 0.0264,
+      "step": 10678
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018309728860611282,
+      "loss": 0.0108,
+      "step": 10679
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018307838074624262,
+      "loss": 0.0146,
+      "step": 10680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018305947233390095,
+      "loss": 0.0233,
+      "step": 10681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018304056336940374,
+      "loss": 0.0119,
+      "step": 10682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001830216538530667,
+      "loss": 0.0062,
+      "step": 10683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018300274378520571,
+      "loss": 0.0037,
+      "step": 10684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018298383316613665,
+      "loss": 0.0362,
+      "step": 10685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001829649219961753,
+      "loss": 0.0234,
+      "step": 10686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001829460102756375,
+      "loss": 0.002,
+      "step": 10687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018292709800483915,
+      "loss": 0.0208,
+      "step": 10688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018290818518409613,
+      "loss": 0.0267,
+      "step": 10689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018288927181372427,
+      "loss": 0.0101,
+      "step": 10690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001828703578940395,
+      "loss": 0.005,
+      "step": 10691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001828514434253577,
+      "loss": 0.0054,
+      "step": 10692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001828325284079948,
+      "loss": 0.0054,
+      "step": 10693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001828136128422667,
+      "loss": 0.0191,
+      "step": 10694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018279469672848929,
+      "loss": 0.0172,
+      "step": 10695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018277578006697855,
+      "loss": 0.0119,
+      "step": 10696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018275686285805043,
+      "loss": 0.007,
+      "step": 10697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018273794510202088,
+      "loss": 0.0039,
+      "step": 10698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018271902679920586,
+      "loss": 0.0093,
+      "step": 10699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018270010794992127,
+      "loss": 0.0104,
+      "step": 10700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018268118855448322,
+      "loss": 0.0092,
+      "step": 10701
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018266226861320757,
+      "loss": 0.0149,
+      "step": 10702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018264334812641045,
+      "loss": 0.0104,
+      "step": 10703
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018262442709440776,
+      "loss": 0.0029,
+      "step": 10704
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001826055055175155,
+      "loss": 0.0165,
+      "step": 10705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001825865833960498,
+      "loss": 0.0038,
+      "step": 10706
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018256766073032661,
+      "loss": 0.0086,
+      "step": 10707
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018254873752066212,
+      "loss": 0.0043,
+      "step": 10708
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018252981376737219,
+      "loss": 0.0083,
+      "step": 10709
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018251088947077298,
+      "loss": 0.0047,
+      "step": 10710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018249196463118055,
+      "loss": 0.0068,
+      "step": 10711
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018247303924891095,
+      "loss": 0.0053,
+      "step": 10712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018245411332428032,
+      "loss": 0.0014,
+      "step": 10713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018243518685760477,
+      "loss": 0.0169,
+      "step": 10714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001824162598492003,
+      "loss": 0.0041,
+      "step": 10715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018239733229938313,
+      "loss": 0.0213,
+      "step": 10716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018237840420846937,
+      "loss": 0.0019,
+      "step": 10717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018235947557677514,
+      "loss": 0.005,
+      "step": 10718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018234054640461656,
+      "loss": 0.0151,
+      "step": 10719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018232161669230984,
+      "loss": 0.0131,
+      "step": 10720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001823026864401711,
+      "loss": 0.0017,
+      "step": 10721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018228375564851654,
+      "loss": 0.0078,
+      "step": 10722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018226482431766232,
+      "loss": 0.005,
+      "step": 10723
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001822458924479246,
+      "loss": 0.004,
+      "step": 10724
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018222696003961962,
+      "loss": 0.0093,
+      "step": 10725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018220802709306355,
+      "loss": 0.0141,
+      "step": 10726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018218909360857266,
+      "loss": 0.0049,
+      "step": 10727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001821701595864631,
+      "loss": 0.0013,
+      "step": 10728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018215122502705122,
+      "loss": 0.0098,
+      "step": 10729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018213228993065313,
+      "loss": 0.003,
+      "step": 10730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018211335429758524,
+      "loss": 0.0088,
+      "step": 10731
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018209441812816363,
+      "loss": 0.0043,
+      "step": 10732
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018207548142270465,
+      "loss": 0.0168,
+      "step": 10733
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018205654418152462,
+      "loss": 0.0222,
+      "step": 10734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018203760640493976,
+      "loss": 0.0024,
+      "step": 10735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001820186680932664,
+      "loss": 0.0097,
+      "step": 10736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018199972924682088,
+      "loss": 0.0118,
+      "step": 10737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018198078986591946,
+      "loss": 0.0056,
+      "step": 10738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001819618499508785,
+      "loss": 0.0029,
+      "step": 10739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018194290950201432,
+      "loss": 0.0057,
+      "step": 10740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018192396851964321,
+      "loss": 0.002,
+      "step": 10741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018190502700408156,
+      "loss": 0.0167,
+      "step": 10742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018188608495564576,
+      "loss": 0.0137,
+      "step": 10743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018186714237465216,
+      "loss": 0.0034,
+      "step": 10744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018184819926141712,
+      "loss": 0.0114,
+      "step": 10745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018182925561625703,
+      "loss": 0.0037,
+      "step": 10746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001818103114394883,
+      "loss": 0.0122,
+      "step": 10747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018179136673142728,
+      "loss": 0.0377,
+      "step": 10748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018177242149239046,
+      "loss": 0.0142,
+      "step": 10749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018175347572269425,
+      "loss": 0.0072,
+      "step": 10750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018173452942265504,
+      "loss": 0.0006,
+      "step": 10751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018171558259258926,
+      "loss": 0.0064,
+      "step": 10752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001816966352328134,
+      "loss": 0.007,
+      "step": 10753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001816776873436439,
+      "loss": 0.0033,
+      "step": 10754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018165873892539726,
+      "loss": 0.0083,
+      "step": 10755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018163978997838987,
+      "loss": 0.0118,
+      "step": 10756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018162084050293828,
+      "loss": 0.0175,
+      "step": 10757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018160189049935892,
+      "loss": 0.0092,
+      "step": 10758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001815829399679684,
+      "loss": 0.0024,
+      "step": 10759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018156398890908316,
+      "loss": 0.0107,
+      "step": 10760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018154503732301967,
+      "loss": 0.0103,
+      "step": 10761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018152608521009456,
+      "loss": 0.0172,
+      "step": 10762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018150713257062434,
+      "loss": 0.0112,
+      "step": 10763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018148817940492548,
+      "loss": 0.0062,
+      "step": 10764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018146922571331462,
+      "loss": 0.0039,
+      "step": 10765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018145027149610828,
+      "loss": 0.0097,
+      "step": 10766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018143131675362302,
+      "loss": 0.0226,
+      "step": 10767
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018141236148617546,
+      "loss": 0.0081,
+      "step": 10768
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018139340569408213,
+      "loss": 0.0034,
+      "step": 10769
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001813744493776597,
+      "loss": 0.0218,
+      "step": 10770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018135549253722472,
+      "loss": 0.0069,
+      "step": 10771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018133653517309387,
+      "loss": 0.0245,
+      "step": 10772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018131757728558371,
+      "loss": 0.0057,
+      "step": 10773
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001812986188750109,
+      "loss": 0.0015,
+      "step": 10774
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018127965994169207,
+      "loss": 0.0037,
+      "step": 10775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018126070048594386,
+      "loss": 0.0062,
+      "step": 10776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018124174050808292,
+      "loss": 0.0017,
+      "step": 10777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000181222780008426,
+      "loss": 0.0127,
+      "step": 10778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001812038189872897,
+      "loss": 0.0029,
+      "step": 10779
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001811848574449907,
+      "loss": 0.0132,
+      "step": 10780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001811658953818457,
+      "loss": 0.0253,
+      "step": 10781
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018114693279817147,
+      "loss": 0.0057,
+      "step": 10782
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018112796969428464,
+      "loss": 0.0067,
+      "step": 10783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018110900607050195,
+      "loss": 0.0049,
+      "step": 10784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018109004192714016,
+      "loss": 0.0134,
+      "step": 10785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018107107726451595,
+      "loss": 0.0045,
+      "step": 10786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018105211208294616,
+      "loss": 0.0078,
+      "step": 10787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018103314638274744,
+      "loss": 0.0183,
+      "step": 10788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018101418016423654,
+      "loss": 0.0051,
+      "step": 10789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018099521342773036,
+      "loss": 0.0048,
+      "step": 10790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001809762461735456,
+      "loss": 0.016,
+      "step": 10791
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018095727840199903,
+      "loss": 0.0065,
+      "step": 10792
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001809383101134075,
+      "loss": 0.0055,
+      "step": 10793
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018091934130808778,
+      "loss": 0.0054,
+      "step": 10794
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001809003719863567,
+      "loss": 0.012,
+      "step": 10795
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018088140214853114,
+      "loss": 0.0102,
+      "step": 10796
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018086243179492778,
+      "loss": 0.0036,
+      "step": 10797
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001808434609258636,
+      "loss": 0.0019,
+      "step": 10798
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018082448954165537,
+      "loss": 0.0066,
+      "step": 10799
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018080551764261999,
+      "loss": 0.0167,
+      "step": 10800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018078654522907432,
+      "loss": 0.0094,
+      "step": 10801
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018076757230133523,
+      "loss": 0.0029,
+      "step": 10802
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001807485988597196,
+      "loss": 0.0104,
+      "step": 10803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001807296249045443,
+      "loss": 0.007,
+      "step": 10804
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018071065043612631,
+      "loss": 0.0062,
+      "step": 10805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001806916754547825,
+      "loss": 0.0052,
+      "step": 10806
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018067269996082972,
+      "loss": 0.0101,
+      "step": 10807
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018065372395458497,
+      "loss": 0.0045,
+      "step": 10808
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018063474743636518,
+      "loss": 0.0063,
+      "step": 10809
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001806157704064873,
+      "loss": 0.0162,
+      "step": 10810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001805967928652682,
+      "loss": 0.0136,
+      "step": 10811
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018057781481302496,
+      "loss": 0.0088,
+      "step": 10812
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018055883625007447,
+      "loss": 0.0156,
+      "step": 10813
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018053985717673374,
+      "loss": 0.0181,
+      "step": 10814
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018052087759331972,
+      "loss": 0.0031,
+      "step": 10815
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018050189750014946,
+      "loss": 0.0026,
+      "step": 10816
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018048291689753992,
+      "loss": 0.0083,
+      "step": 10817
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018046393578580812,
+      "loss": 0.0099,
+      "step": 10818
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001804449541652711,
+      "loss": 0.0155,
+      "step": 10819
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018042597203624588,
+      "loss": 0.0033,
+      "step": 10820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018040698939904949,
+      "loss": 0.0144,
+      "step": 10821
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018038800625399897,
+      "loss": 0.0167,
+      "step": 10822
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001803690226014114,
+      "loss": 0.0178,
+      "step": 10823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018035003844160384,
+      "loss": 0.0013,
+      "step": 10824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018033105377489333,
+      "loss": 0.0202,
+      "step": 10825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018031206860159697,
+      "loss": 0.0137,
+      "step": 10826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018029308292203183,
+      "loss": 0.0048,
+      "step": 10827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018027409673651508,
+      "loss": 0.0077,
+      "step": 10828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018025511004536373,
+      "loss": 0.0094,
+      "step": 10829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001802361228488949,
+      "loss": 0.0171,
+      "step": 10830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018021713514742576,
+      "loss": 0.0049,
+      "step": 10831
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018019814694127342,
+      "loss": 0.0072,
+      "step": 10832
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001801791582307551,
+      "loss": 0.0099,
+      "step": 10833
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018016016901618777,
+      "loss": 0.008,
+      "step": 10834
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018014117929788875,
+      "loss": 0.0079,
+      "step": 10835
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001801221890761751,
+      "loss": 0.0199,
+      "step": 10836
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001801031983513641,
+      "loss": 0.0171,
+      "step": 10837
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018008420712377276,
+      "loss": 0.0014,
+      "step": 10838
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018006521539371844,
+      "loss": 0.0153,
+      "step": 10839
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018004622316151824,
+      "loss": 0.0013,
+      "step": 10840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001800272304274894,
+      "loss": 0.0183,
+      "step": 10841
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018000823719194913,
+      "loss": 0.0091,
+      "step": 10842
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001799892434552147,
+      "loss": 0.0155,
+      "step": 10843
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017997024921760326,
+      "loss": 0.0032,
+      "step": 10844
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017995125447943207,
+      "loss": 0.0302,
+      "step": 10845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001799322592410184,
+      "loss": 0.0074,
+      "step": 10846
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017991326350267955,
+      "loss": 0.0016,
+      "step": 10847
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001798942672647327,
+      "loss": 0.0058,
+      "step": 10848
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017987527052749518,
+      "loss": 0.0033,
+      "step": 10849
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001798562732912842,
+      "loss": 0.0033,
+      "step": 10850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001798372755564171,
+      "loss": 0.0539,
+      "step": 10851
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001798182773232112,
+      "loss": 0.0038,
+      "step": 10852
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017979927859198378,
+      "loss": 0.0138,
+      "step": 10853
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017978027936305216,
+      "loss": 0.0029,
+      "step": 10854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017976127963673365,
+      "loss": 0.0079,
+      "step": 10855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001797422794133456,
+      "loss": 0.0144,
+      "step": 10856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017972327869320538,
+      "loss": 0.0045,
+      "step": 10857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017970427747663025,
+      "loss": 0.0157,
+      "step": 10858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001796852757639376,
+      "loss": 0.0031,
+      "step": 10859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017966627355544485,
+      "loss": 0.0198,
+      "step": 10860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017964727085146933,
+      "loss": 0.0364,
+      "step": 10861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017962826765232837,
+      "loss": 0.0051,
+      "step": 10862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017960926395833946,
+      "loss": 0.014,
+      "step": 10863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017959025976981995,
+      "loss": 0.0091,
+      "step": 10864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017957125508708726,
+      "loss": 0.0362,
+      "step": 10865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017955224991045876,
+      "loss": 0.0062,
+      "step": 10866
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017953324424025191,
+      "loss": 0.0046,
+      "step": 10867
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017951423807678415,
+      "loss": 0.0104,
+      "step": 10868
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017949523142037286,
+      "loss": 0.0093,
+      "step": 10869
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001794762242713356,
+      "loss": 0.0164,
+      "step": 10870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001794572166299897,
+      "loss": 0.0032,
+      "step": 10871
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017943820849665267,
+      "loss": 0.0068,
+      "step": 10872
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000179419199871642,
+      "loss": 0.0287,
+      "step": 10873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017940019075527518,
+      "loss": 0.0016,
+      "step": 10874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001793811811478697,
+      "loss": 0.0088,
+      "step": 10875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017936217104974297,
+      "loss": 0.0113,
+      "step": 10876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001793431604612126,
+      "loss": 0.0102,
+      "step": 10877
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017932414938259605,
+      "loss": 0.0099,
+      "step": 10878
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017930513781421085,
+      "loss": 0.027,
+      "step": 10879
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017928612575637454,
+      "loss": 0.0047,
+      "step": 10880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017926711320940465,
+      "loss": 0.0151,
+      "step": 10881
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001792481001736187,
+      "loss": 0.0014,
+      "step": 10882
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017922908664933426,
+      "loss": 0.0045,
+      "step": 10883
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017921007263686895,
+      "loss": 0.019,
+      "step": 10884
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017919105813654025,
+      "loss": 0.008,
+      "step": 10885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017917204314866578,
+      "loss": 0.0069,
+      "step": 10886
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017915302767356313,
+      "loss": 0.0109,
+      "step": 10887
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017913401171154993,
+      "loss": 0.017,
+      "step": 10888
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001791149952629437,
+      "loss": 0.0059,
+      "step": 10889
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017909597832806208,
+      "loss": 0.0065,
+      "step": 10890
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001790769609072227,
+      "loss": 0.0072,
+      "step": 10891
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017905794300074318,
+      "loss": 0.0101,
+      "step": 10892
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017903892460894118,
+      "loss": 0.0102,
+      "step": 10893
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017901990573213433,
+      "loss": 0.0026,
+      "step": 10894
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017900088637064028,
+      "loss": 0.0012,
+      "step": 10895
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017898186652477667,
+      "loss": 0.0168,
+      "step": 10896
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017896284619486116,
+      "loss": 0.0102,
+      "step": 10897
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001789438253812115,
+      "loss": 0.0166,
+      "step": 10898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017892480408414528,
+      "loss": 0.0089,
+      "step": 10899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017890578230398023,
+      "loss": 0.0142,
+      "step": 10900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017888676004103408,
+      "loss": 0.0104,
+      "step": 10901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017886773729562446,
+      "loss": 0.0073,
+      "step": 10902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001788487140680692,
+      "loss": 0.0131,
+      "step": 10903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001788296903586859,
+      "loss": 0.0009,
+      "step": 10904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001788106661677924,
+      "loss": 0.0215,
+      "step": 10905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017879164149570638,
+      "loss": 0.0075,
+      "step": 10906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001787726163427456,
+      "loss": 0.0035,
+      "step": 10907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001787535907092278,
+      "loss": 0.0078,
+      "step": 10908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017873456459547074,
+      "loss": 0.0054,
+      "step": 10909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017871553800179223,
+      "loss": 0.0084,
+      "step": 10910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017869651092851008,
+      "loss": 0.011,
+      "step": 10911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017867748337594197,
+      "loss": 0.011,
+      "step": 10912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017865845534440576,
+      "loss": 0.0027,
+      "step": 10913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017863942683421925,
+      "loss": 0.0193,
+      "step": 10914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017862039784570025,
+      "loss": 0.0209,
+      "step": 10915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001786013683791666,
+      "loss": 0.0042,
+      "step": 10916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017858233843493613,
+      "loss": 0.004,
+      "step": 10917
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017856330801332661,
+      "loss": 0.0076,
+      "step": 10918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017854427711465595,
+      "loss": 0.0112,
+      "step": 10919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017852524573924196,
+      "loss": 0.0049,
+      "step": 10920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017850621388740257,
+      "loss": 0.0007,
+      "step": 10921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017848718155945555,
+      "loss": 0.0057,
+      "step": 10922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017846814875571886,
+      "loss": 0.0138,
+      "step": 10923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017844911547651034,
+      "loss": 0.0145,
+      "step": 10924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017843008172214783,
+      "loss": 0.0064,
+      "step": 10925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017841104749294936,
+      "loss": 0.0182,
+      "step": 10926
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017839201278923274,
+      "loss": 0.0056,
+      "step": 10927
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001783729776113159,
+      "loss": 0.0032,
+      "step": 10928
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001783539419595168,
+      "loss": 0.0121,
+      "step": 10929
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017833490583415336,
+      "loss": 0.0047,
+      "step": 10930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001783158692355435,
+      "loss": 0.0089,
+      "step": 10931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017829683216400515,
+      "loss": 0.0025,
+      "step": 10932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017827779461985628,
+      "loss": 0.0046,
+      "step": 10933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017825875660341492,
+      "loss": 0.01,
+      "step": 10934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017823971811499893,
+      "loss": 0.0167,
+      "step": 10935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017822067915492635,
+      "loss": 0.0065,
+      "step": 10936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017820163972351514,
+      "loss": 0.004,
+      "step": 10937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017818259982108336,
+      "loss": 0.0098,
+      "step": 10938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017816355944794897,
+      "loss": 0.0079,
+      "step": 10939
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017814451860442994,
+      "loss": 0.0031,
+      "step": 10940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001781254772908443,
+      "loss": 0.0013,
+      "step": 10941
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017810643550751016,
+      "loss": 0.0081,
+      "step": 10942
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017808739325474545,
+      "loss": 0.0007,
+      "step": 10943
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017806835053286825,
+      "loss": 0.0251,
+      "step": 10944
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017804930734219662,
+      "loss": 0.0112,
+      "step": 10945
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017803026368304862,
+      "loss": 0.0099,
+      "step": 10946
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001780112195557423,
+      "loss": 0.0017,
+      "step": 10947
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017799217496059576,
+      "loss": 0.0073,
+      "step": 10948
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017797312989792705,
+      "loss": 0.0032,
+      "step": 10949
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017795408436805423,
+      "loss": 0.0134,
+      "step": 10950
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017793503837129548,
+      "loss": 0.0149,
+      "step": 10951
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017791599190796883,
+      "loss": 0.0088,
+      "step": 10952
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017789694497839244,
+      "loss": 0.0048,
+      "step": 10953
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001778778975828844,
+      "loss": 0.0064,
+      "step": 10954
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017785884972176282,
+      "loss": 0.0128,
+      "step": 10955
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001778398013953459,
+      "loss": 0.0051,
+      "step": 10956
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001778207526039517,
+      "loss": 0.0075,
+      "step": 10957
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001778017033478985,
+      "loss": 0.0092,
+      "step": 10958
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017778265362750433,
+      "loss": 0.0064,
+      "step": 10959
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017776360344308738,
+      "loss": 0.0269,
+      "step": 10960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017774455279496586,
+      "loss": 0.0099,
+      "step": 10961
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017772550168345798,
+      "loss": 0.0062,
+      "step": 10962
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017770645010888183,
+      "loss": 0.0152,
+      "step": 10963
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017768739807155567,
+      "loss": 0.0177,
+      "step": 10964
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001776683455717977,
+      "loss": 0.0051,
+      "step": 10965
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001776492926099261,
+      "loss": 0.0025,
+      "step": 10966
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001776302391862592,
+      "loss": 0.0037,
+      "step": 10967
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001776111853011151,
+      "loss": 0.004,
+      "step": 10968
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001775921309548121,
+      "loss": 0.0107,
+      "step": 10969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001775730761476684,
+      "loss": 0.0126,
+      "step": 10970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017755402088000233,
+      "loss": 0.0065,
+      "step": 10971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001775349651521321,
+      "loss": 0.0034,
+      "step": 10972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017751590896437596,
+      "loss": 0.0151,
+      "step": 10973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017749685231705216,
+      "loss": 0.0106,
+      "step": 10974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017747779521047906,
+      "loss": 0.0037,
+      "step": 10975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001774587376449749,
+      "loss": 0.0037,
+      "step": 10976
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017743967962085798,
+      "loss": 0.004,
+      "step": 10977
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017742062113844663,
+      "loss": 0.0006,
+      "step": 10978
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017740156219805915,
+      "loss": 0.0194,
+      "step": 10979
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017738250280001383,
+      "loss": 0.0108,
+      "step": 10980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001773634429446291,
+      "loss": 0.0148,
+      "step": 10981
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017734438263222318,
+      "loss": 0.0099,
+      "step": 10982
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001773253218631144,
+      "loss": 0.0253,
+      "step": 10983
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017730626063762123,
+      "loss": 0.0058,
+      "step": 10984
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017728719895606192,
+      "loss": 0.0037,
+      "step": 10985
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001772681368187549,
+      "loss": 0.0055,
+      "step": 10986
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017724907422601852,
+      "loss": 0.0109,
+      "step": 10987
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017723001117817116,
+      "loss": 0.0201,
+      "step": 10988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001772109476755312,
+      "loss": 0.0123,
+      "step": 10989
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771918837184171,
+      "loss": 0.0156,
+      "step": 10990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771728193071472,
+      "loss": 0.023,
+      "step": 10991
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771537544420399,
+      "loss": 0.0038,
+      "step": 10992
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017713468912341365,
+      "loss": 0.0059,
+      "step": 10993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771156233515869,
+      "loss": 0.0029,
+      "step": 10994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017709655712687804,
+      "loss": 0.0091,
+      "step": 10995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017707749044960547,
+      "loss": 0.0023,
+      "step": 10996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017705842332008774,
+      "loss": 0.003,
+      "step": 10997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017703935573864328,
+      "loss": 0.0157,
+      "step": 10998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017702028770559053,
+      "loss": 0.0095,
+      "step": 10999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017700121922124796,
+      "loss": 0.0292,
+      "step": 11000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017698215028593404,
+      "loss": 0.0019,
+      "step": 11001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017696308089996731,
+      "loss": 0.0028,
+      "step": 11002
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017694401106366626,
+      "loss": 0.0123,
+      "step": 11003
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017692494077734932,
+      "loss": 0.0074,
+      "step": 11004
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001769058700413351,
+      "loss": 0.0116,
+      "step": 11005
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000176886798855942,
+      "loss": 0.0117,
+      "step": 11006
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017686772722148863,
+      "loss": 0.0061,
+      "step": 11007
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017684865513829345,
+      "loss": 0.0068,
+      "step": 11008
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017682958260667513,
+      "loss": 0.0081,
+      "step": 11009
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001768105096269521,
+      "loss": 0.0136,
+      "step": 11010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017679143619944295,
+      "loss": 0.0052,
+      "step": 11011
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017677236232446625,
+      "loss": 0.0099,
+      "step": 11012
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017675328800234057,
+      "loss": 0.0076,
+      "step": 11013
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017673421323338446,
+      "loss": 0.0177,
+      "step": 11014
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017671513801791654,
+      "loss": 0.0098,
+      "step": 11015
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017669606235625536,
+      "loss": 0.0134,
+      "step": 11016
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001766769862487196,
+      "loss": 0.0045,
+      "step": 11017
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017665790969562773,
+      "loss": 0.0243,
+      "step": 11018
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001766388326972985,
+      "loss": 0.0121,
+      "step": 11019
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017661975525405047,
+      "loss": 0.0111,
+      "step": 11020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017660067736620228,
+      "loss": 0.0066,
+      "step": 11021
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001765815990340726,
+      "loss": 0.0066,
+      "step": 11022
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017656252025798,
+      "loss": 0.0057,
+      "step": 11023
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017654344103824318,
+      "loss": 0.003,
+      "step": 11024
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001765243613751808,
+      "loss": 0.0226,
+      "step": 11025
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001765052812691115,
+      "loss": 0.007,
+      "step": 11026
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017648620072035397,
+      "loss": 0.0033,
+      "step": 11027
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001764671197292269,
+      "loss": 0.0124,
+      "step": 11028
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017644803829604895,
+      "loss": 0.0062,
+      "step": 11029
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001764289564211389,
+      "loss": 0.0142,
+      "step": 11030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001764098741048153,
+      "loss": 0.0265,
+      "step": 11031
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017639079134739702,
+      "loss": 0.0034,
+      "step": 11032
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001763717081492027,
+      "loss": 0.002,
+      "step": 11033
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017635262451055103,
+      "loss": 0.0068,
+      "step": 11034
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017633354043176085,
+      "loss": 0.0135,
+      "step": 11035
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001763144559131508,
+      "loss": 0.0227,
+      "step": 11036
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017629537095503967,
+      "loss": 0.0353,
+      "step": 11037
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017627628555774618,
+      "loss": 0.0022,
+      "step": 11038
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017625719972158915,
+      "loss": 0.0043,
+      "step": 11039
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001762381134468873,
+      "loss": 0.0111,
+      "step": 11040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017621902673395948,
+      "loss": 0.01,
+      "step": 11041
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017619993958312437,
+      "loss": 0.0008,
+      "step": 11042
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017618085199470082,
+      "loss": 0.0067,
+      "step": 11043
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017616176396900764,
+      "loss": 0.0249,
+      "step": 11044
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017614267550636358,
+      "loss": 0.0169,
+      "step": 11045
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017612358660708757,
+      "loss": 0.0039,
+      "step": 11046
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017610449727149832,
+      "loss": 0.0092,
+      "step": 11047
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017608540749991465,
+      "loss": 0.018,
+      "step": 11048
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017606631729265542,
+      "loss": 0.0044,
+      "step": 11049
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017604722665003956,
+      "loss": 0.0065,
+      "step": 11050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017602813557238579,
+      "loss": 0.0188,
+      "step": 11051
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017600904406001305,
+      "loss": 0.025,
+      "step": 11052
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017598995211324015,
+      "loss": 0.0103,
+      "step": 11053
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017597085973238604,
+      "loss": 0.0095,
+      "step": 11054
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017595176691776956,
+      "loss": 0.0164,
+      "step": 11055
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017593267366970952,
+      "loss": 0.0011,
+      "step": 11056
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017591357998852493,
+      "loss": 0.0113,
+      "step": 11057
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001758944858745346,
+      "loss": 0.0123,
+      "step": 11058
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001758753913280575,
+      "loss": 0.0031,
+      "step": 11059
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001758562963494125,
+      "loss": 0.0142,
+      "step": 11060
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017583720093891856,
+      "loss": 0.0029,
+      "step": 11061
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017581810509689457,
+      "loss": 0.0182,
+      "step": 11062
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001757990088236595,
+      "loss": 0.0145,
+      "step": 11063
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017577991211953234,
+      "loss": 0.0073,
+      "step": 11064
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017576081498483194,
+      "loss": 0.0013,
+      "step": 11065
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017574171741987729,
+      "loss": 0.0091,
+      "step": 11066
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017572261942498736,
+      "loss": 0.0156,
+      "step": 11067
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017570352100048116,
+      "loss": 0.0066,
+      "step": 11068
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017568442214667762,
+      "loss": 0.0076,
+      "step": 11069
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017566532286389578,
+      "loss": 0.0124,
+      "step": 11070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017564622315245458,
+      "loss": 0.0019,
+      "step": 11071
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017562712301267302,
+      "loss": 0.0103,
+      "step": 11072
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017560802244487018,
+      "loss": 0.0109,
+      "step": 11073
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017558892144936498,
+      "loss": 0.01,
+      "step": 11074
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001755698200264765,
+      "loss": 0.0238,
+      "step": 11075
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001755507181765238,
+      "loss": 0.01,
+      "step": 11076
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017553161589982586,
+      "loss": 0.0019,
+      "step": 11077
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017551251319670172,
+      "loss": 0.0081,
+      "step": 11078
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017549341006747047,
+      "loss": 0.0082,
+      "step": 11079
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001754743065124511,
+      "loss": 0.0026,
+      "step": 11080
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017545520253196275,
+      "loss": 0.0438,
+      "step": 11081
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017543609812632448,
+      "loss": 0.0126,
+      "step": 11082
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001754169932958554,
+      "loss": 0.0027,
+      "step": 11083
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017539788804087447,
+      "loss": 0.0058,
+      "step": 11084
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017537878236170094,
+      "loss": 0.0114,
+      "step": 11085
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001753596762586538,
+      "loss": 0.0231,
+      "step": 11086
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017534056973205224,
+      "loss": 0.0091,
+      "step": 11087
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017532146278221528,
+      "loss": 0.01,
+      "step": 11088
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001753023554094621,
+      "loss": 0.004,
+      "step": 11089
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017528324761411184,
+      "loss": 0.0164,
+      "step": 11090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017526413939648354,
+      "loss": 0.0148,
+      "step": 11091
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017524503075689652,
+      "loss": 0.0101,
+      "step": 11092
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017522592169566984,
+      "loss": 0.0139,
+      "step": 11093
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001752068122131226,
+      "loss": 0.0107,
+      "step": 11094
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000175187702309574,
+      "loss": 0.0063,
+      "step": 11095
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017516859198534327,
+      "loss": 0.0107,
+      "step": 11096
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017514948124074955,
+      "loss": 0.0116,
+      "step": 11097
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000175130370076112,
+      "loss": 0.02,
+      "step": 11098
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017511125849174985,
+      "loss": 0.0111,
+      "step": 11099
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017509214648798226,
+      "loss": 0.0158,
+      "step": 11100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017507303406512847,
+      "loss": 0.0039,
+      "step": 11101
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017505392122350768,
+      "loss": 0.0005,
+      "step": 11102
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001750348079634391,
+      "loss": 0.0031,
+      "step": 11103
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000175015694285242,
+      "loss": 0.0052,
+      "step": 11104
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017499658018923557,
+      "loss": 0.0107,
+      "step": 11105
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001749774656757391,
+      "loss": 0.0224,
+      "step": 11106
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017495835074507173,
+      "loss": 0.0176,
+      "step": 11107
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017493923539755284,
+      "loss": 0.0075,
+      "step": 11108
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001749201196335016,
+      "loss": 0.0234,
+      "step": 11109
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017490100345323738,
+      "loss": 0.0287,
+      "step": 11110
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017488188685707936,
+      "loss": 0.009,
+      "step": 11111
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017486276984534686,
+      "loss": 0.0103,
+      "step": 11112
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001748436524183592,
+      "loss": 0.0014,
+      "step": 11113
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017482453457643563,
+      "loss": 0.0085,
+      "step": 11114
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001748054163198955,
+      "loss": 0.0061,
+      "step": 11115
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017478629764905806,
+      "loss": 0.005,
+      "step": 11116
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001747671785642427,
+      "loss": 0.0039,
+      "step": 11117
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017474805906576865,
+      "loss": 0.0063,
+      "step": 11118
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001747289391539553,
+      "loss": 0.0161,
+      "step": 11119
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001747098188291221,
+      "loss": 0.0161,
+      "step": 11120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017469069809158814,
+      "loss": 0.0027,
+      "step": 11121
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017467157694167298,
+      "loss": 0.0118,
+      "step": 11122
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001746524553796959,
+      "loss": 0.0211,
+      "step": 11123
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017463333340597632,
+      "loss": 0.0043,
+      "step": 11124
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017461421102083353,
+      "loss": 0.0069,
+      "step": 11125
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017459508822458694,
+      "loss": 0.007,
+      "step": 11126
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017457596501755597,
+      "loss": 0.0122,
+      "step": 11127
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017455684140006,
+      "loss": 0.0061,
+      "step": 11128
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017453771737241847,
+      "loss": 0.0087,
+      "step": 11129
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017451859293495068,
+      "loss": 0.0059,
+      "step": 11130
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017449946808797612,
+      "loss": 0.0092,
+      "step": 11131
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001744803428318142,
+      "loss": 0.0011,
+      "step": 11132
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001744612171667843,
+      "loss": 0.0032,
+      "step": 11133
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017444209109320597,
+      "loss": 0.0114,
+      "step": 11134
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017442296461139857,
+      "loss": 0.0087,
+      "step": 11135
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001744038377216815,
+      "loss": 0.0057,
+      "step": 11136
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001743847104243743,
+      "loss": 0.0044,
+      "step": 11137
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001743655827197965,
+      "loss": 0.0121,
+      "step": 11138
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017434645460826737,
+      "loss": 0.0091,
+      "step": 11139
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017432732609010652,
+      "loss": 0.0079,
+      "step": 11140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001743081971656334,
+      "loss": 0.0054,
+      "step": 11141
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017428906783516752,
+      "loss": 0.0033,
+      "step": 11142
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017426993809902832,
+      "loss": 0.0054,
+      "step": 11143
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017425080795753536,
+      "loss": 0.0084,
+      "step": 11144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017423167741100815,
+      "loss": 0.0068,
+      "step": 11145
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017421254645976614,
+      "loss": 0.0055,
+      "step": 11146
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017419341510412897,
+      "loss": 0.0053,
+      "step": 11147
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017417428334441605,
+      "loss": 0.0029,
+      "step": 11148
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000174155151180947,
+      "loss": 0.0018,
+      "step": 11149
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017413601861404127,
+      "loss": 0.013,
+      "step": 11150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001741168856440185,
+      "loss": 0.0085,
+      "step": 11151
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001740977522711982,
+      "loss": 0.0169,
+      "step": 11152
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017407861849589999,
+      "loss": 0.0262,
+      "step": 11153
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017405948431844338,
+      "loss": 0.0151,
+      "step": 11154
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017404034973914794,
+      "loss": 0.0074,
+      "step": 11155
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017402121475833333,
+      "loss": 0.0023,
+      "step": 11156
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017400207937631908,
+      "loss": 0.0058,
+      "step": 11157
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001739829435934248,
+      "loss": 0.0115,
+      "step": 11158
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017396380740997003,
+      "loss": 0.0039,
+      "step": 11159
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001739446708262745,
+      "loss": 0.0077,
+      "step": 11160
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017392553384265778,
+      "loss": 0.0121,
+      "step": 11161
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017390639645943945,
+      "loss": 0.0101,
+      "step": 11162
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017388725867693915,
+      "loss": 0.01,
+      "step": 11163
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017386812049547656,
+      "loss": 0.003,
+      "step": 11164
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017384898191537132,
+      "loss": 0.0089,
+      "step": 11165
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017382984293694306,
+      "loss": 0.0094,
+      "step": 11166
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017381070356051144,
+      "loss": 0.0145,
+      "step": 11167
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017379156378639614,
+      "loss": 0.0221,
+      "step": 11168
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001737724236149168,
+      "loss": 0.007,
+      "step": 11169
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001737532830463931,
+      "loss": 0.0111,
+      "step": 11170
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017373414208114477,
+      "loss": 0.0033,
+      "step": 11171
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017371500071949143,
+      "loss": 0.0118,
+      "step": 11172
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001736958589617528,
+      "loss": 0.0147,
+      "step": 11173
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001736767168082486,
+      "loss": 0.0048,
+      "step": 11174
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017365757425929858,
+      "loss": 0.0101,
+      "step": 11175
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017363843131522238,
+      "loss": 0.0155,
+      "step": 11176
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017361928797633976,
+      "loss": 0.0168,
+      "step": 11177
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017360014424297045,
+      "loss": 0.001,
+      "step": 11178
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017358100011543418,
+      "loss": 0.0048,
+      "step": 11179
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017356185559405074,
+      "loss": 0.0045,
+      "step": 11180
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017354271067913978,
+      "loss": 0.0048,
+      "step": 11181
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001735235653710211,
+      "loss": 0.0096,
+      "step": 11182
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017350441967001453,
+      "loss": 0.0131,
+      "step": 11183
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017348527357643974,
+      "loss": 0.0048,
+      "step": 11184
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017346612709061656,
+      "loss": 0.0088,
+      "step": 11185
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001734469802128648,
+      "loss": 0.0012,
+      "step": 11186
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017342783294350416,
+      "loss": 0.0047,
+      "step": 11187
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001734086852828545,
+      "loss": 0.001,
+      "step": 11188
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017338953723123563,
+      "loss": 0.0169,
+      "step": 11189
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001733703887889673,
+      "loss": 0.0035,
+      "step": 11190
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001733512399563694,
+      "loss": 0.0038,
+      "step": 11191
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017333209073376168,
+      "loss": 0.0006,
+      "step": 11192
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017331294112146403,
+      "loss": 0.01,
+      "step": 11193
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017329379111979626,
+      "loss": 0.0129,
+      "step": 11194
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017327464072907817,
+      "loss": 0.0116,
+      "step": 11195
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017325548994962965,
+      "loss": 0.0061,
+      "step": 11196
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017323633878177057,
+      "loss": 0.003,
+      "step": 11197
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001732171872258208,
+      "loss": 0.0155,
+      "step": 11198
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017319803528210016,
+      "loss": 0.0143,
+      "step": 11199
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017317888295092853,
+      "loss": 0.0039,
+      "step": 11200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001731597302326258,
+      "loss": 0.0305,
+      "step": 11201
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017314057712751184,
+      "loss": 0.0087,
+      "step": 11202
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017312142363590659,
+      "loss": 0.001,
+      "step": 11203
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017310226975812988,
+      "loss": 0.0078,
+      "step": 11204
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001730831154945017,
+      "loss": 0.0032,
+      "step": 11205
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001730639608453419,
+      "loss": 0.0118,
+      "step": 11206
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017304480581097043,
+      "loss": 0.0024,
+      "step": 11207
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001730256503917072,
+      "loss": 0.0093,
+      "step": 11208
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017300649458787214,
+      "loss": 0.0399,
+      "step": 11209
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017298733839978518,
+      "loss": 0.0058,
+      "step": 11210
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017296818182776632,
+      "loss": 0.0105,
+      "step": 11211
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017294902487213548,
+      "loss": 0.0139,
+      "step": 11212
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017292986753321253,
+      "loss": 0.0115,
+      "step": 11213
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017291070981131755,
+      "loss": 0.0141,
+      "step": 11214
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017289155170677043,
+      "loss": 0.005,
+      "step": 11215
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017287239321989119,
+      "loss": 0.0062,
+      "step": 11216
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001728532343509999,
+      "loss": 0.0039,
+      "step": 11217
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017283407510041638,
+      "loss": 0.0046,
+      "step": 11218
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001728149154684607,
+      "loss": 0.0103,
+      "step": 11219
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017279575545545286,
+      "loss": 0.0069,
+      "step": 11220
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017277659506171296,
+      "loss": 0.0022,
+      "step": 11221
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017275743428756083,
+      "loss": 0.01,
+      "step": 11222
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017273827313331664,
+      "loss": 0.0095,
+      "step": 11223
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017271911159930036,
+      "loss": 0.0259,
+      "step": 11224
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017269994968583202,
+      "loss": 0.0072,
+      "step": 11225
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017268078739323168,
+      "loss": 0.0019,
+      "step": 11226
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017266162472181934,
+      "loss": 0.0037,
+      "step": 11227
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017264246167191514,
+      "loss": 0.0183,
+      "step": 11228
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017262329824383903,
+      "loss": 0.0042,
+      "step": 11229
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001726041344379112,
+      "loss": 0.0171,
+      "step": 11230
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017258497025445164,
+      "loss": 0.0194,
+      "step": 11231
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017256580569378045,
+      "loss": 0.0086,
+      "step": 11232
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017254664075621768,
+      "loss": 0.0055,
+      "step": 11233
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017252747544208347,
+      "loss": 0.0058,
+      "step": 11234
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017250830975169788,
+      "loss": 0.0077,
+      "step": 11235
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017248914368538107,
+      "loss": 0.0162,
+      "step": 11236
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017246997724345306,
+      "loss": 0.0055,
+      "step": 11237
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017245081042623403,
+      "loss": 0.0124,
+      "step": 11238
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001724316432340441,
+      "loss": 0.0048,
+      "step": 11239
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001724124756672034,
+      "loss": 0.0095,
+      "step": 11240
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017239330772603204,
+      "loss": 0.0052,
+      "step": 11241
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017237413941085015,
+      "loss": 0.0116,
+      "step": 11242
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001723549707219779,
+      "loss": 0.0048,
+      "step": 11243
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017233580165973542,
+      "loss": 0.0026,
+      "step": 11244
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017231663222444296,
+      "loss": 0.0159,
+      "step": 11245
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017229746241642052,
+      "loss": 0.0044,
+      "step": 11246
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001722782922359884,
+      "loss": 0.012,
+      "step": 11247
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001722591216834668,
+      "loss": 0.0047,
+      "step": 11248
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017223995075917584,
+      "loss": 0.0059,
+      "step": 11249
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001722207794634357,
+      "loss": 0.0101,
+      "step": 11250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017220160779656658,
+      "loss": 0.0139,
+      "step": 11251
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017218243575888872,
+      "loss": 0.0038,
+      "step": 11252
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017216326335072233,
+      "loss": 0.0174,
+      "step": 11253
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017214409057238764,
+      "loss": 0.0028,
+      "step": 11254
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017212491742420475,
+      "loss": 0.0084,
+      "step": 11255
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017210574390649403,
+      "loss": 0.0099,
+      "step": 11256
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001720865700195756,
+      "loss": 0.0332,
+      "step": 11257
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017206739576376982,
+      "loss": 0.0085,
+      "step": 11258
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017204822113939684,
+      "loss": 0.0116,
+      "step": 11259
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017202904614677694,
+      "loss": 0.0041,
+      "step": 11260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001720098707862304,
+      "loss": 0.0079,
+      "step": 11261
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017199069505807748,
+      "loss": 0.0097,
+      "step": 11262
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001719715189626385,
+      "loss": 0.003,
+      "step": 11263
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017195234250023361,
+      "loss": 0.0025,
+      "step": 11264
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017193316567118316,
+      "loss": 0.0054,
+      "step": 11265
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017191398847580746,
+      "loss": 0.0021,
+      "step": 11266
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017189481091442677,
+      "loss": 0.0123,
+      "step": 11267
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017187563298736144,
+      "loss": 0.0083,
+      "step": 11268
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017185645469493168,
+      "loss": 0.0153,
+      "step": 11269
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001718372760374579,
+      "loss": 0.0043,
+      "step": 11270
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017181809701526042,
+      "loss": 0.0042,
+      "step": 11271
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017179891762865955,
+      "loss": 0.0092,
+      "step": 11272
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001717797378779756,
+      "loss": 0.0178,
+      "step": 11273
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017176055776352888,
+      "loss": 0.0011,
+      "step": 11274
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001717413772856398,
+      "loss": 0.0084,
+      "step": 11275
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017172219644462862,
+      "loss": 0.0056,
+      "step": 11276
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017170301524081582,
+      "loss": 0.0109,
+      "step": 11277
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017168383367452167,
+      "loss": 0.0059,
+      "step": 11278
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017166465174606654,
+      "loss": 0.0089,
+      "step": 11279
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001716454694557709,
+      "loss": 0.0524,
+      "step": 11280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000171626286803955,
+      "loss": 0.0093,
+      "step": 11281
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017160710379093933,
+      "loss": 0.0055,
+      "step": 11282
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017158792041704424,
+      "loss": 0.0014,
+      "step": 11283
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001715687366825901,
+      "loss": 0.0131,
+      "step": 11284
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017154955258789735,
+      "loss": 0.0024,
+      "step": 11285
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017153036813328642,
+      "loss": 0.0046,
+      "step": 11286
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017151118331907767,
+      "loss": 0.0088,
+      "step": 11287
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017149199814559153,
+      "loss": 0.0157,
+      "step": 11288
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001714728126131485,
+      "loss": 0.0037,
+      "step": 11289
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017145362672206895,
+      "loss": 0.0065,
+      "step": 11290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017143444047267335,
+      "loss": 0.0203,
+      "step": 11291
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017141525386528213,
+      "loss": 0.0148,
+      "step": 11292
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017139606690021574,
+      "loss": 0.0052,
+      "step": 11293
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017137687957779463,
+      "loss": 0.0099,
+      "step": 11294
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017135769189833934,
+      "loss": 0.0128,
+      "step": 11295
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001713385038621702,
+      "loss": 0.0027,
+      "step": 11296
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001713193154696078,
+      "loss": 0.0187,
+      "step": 11297
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001713001267209726,
+      "loss": 0.0022,
+      "step": 11298
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000171280937616585,
+      "loss": 0.0076,
+      "step": 11299
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017126174815676568,
+      "loss": 0.0109,
+      "step": 11300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017124255834183498,
+      "loss": 0.0136,
+      "step": 11301
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017122336817211345,
+      "loss": 0.0067,
+      "step": 11302
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017120417764792165,
+      "loss": 0.0126,
+      "step": 11303
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017118498676958003,
+      "loss": 0.003,
+      "step": 11304
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017116579553740917,
+      "loss": 0.0041,
+      "step": 11305
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017114660395172953,
+      "loss": 0.0116,
+      "step": 11306
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001711274120128617,
+      "loss": 0.0085,
+      "step": 11307
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001711082197211262,
+      "loss": 0.0036,
+      "step": 11308
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001710890270768436,
+      "loss": 0.0084,
+      "step": 11309
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017106983408033443,
+      "loss": 0.0017,
+      "step": 11310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017105064073191925,
+      "loss": 0.0248,
+      "step": 11311
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017103144703191865,
+      "loss": 0.0037,
+      "step": 11312
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017101225298065317,
+      "loss": 0.0097,
+      "step": 11313
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017099305857844345,
+      "loss": 0.0094,
+      "step": 11314
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017097386382561,
+      "loss": 0.0086,
+      "step": 11315
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017095466872247344,
+      "loss": 0.002,
+      "step": 11316
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001709354732693543,
+      "loss": 0.0192,
+      "step": 11317
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001709162774665733,
+      "loss": 0.0153,
+      "step": 11318
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017089708131445093,
+      "loss": 0.007,
+      "step": 11319
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017087788481330788,
+      "loss": 0.0294,
+      "step": 11320
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017085868796346476,
+      "loss": 0.0043,
+      "step": 11321
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017083949076524216,
+      "loss": 0.0064,
+      "step": 11322
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017082029321896077,
+      "loss": 0.0106,
+      "step": 11323
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017080109532494113,
+      "loss": 0.0063,
+      "step": 11324
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017078189708350397,
+      "loss": 0.0081,
+      "step": 11325
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017076269849496987,
+      "loss": 0.0057,
+      "step": 11326
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017074349955965954,
+      "loss": 0.0087,
+      "step": 11327
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017072430027789362,
+      "loss": 0.0119,
+      "step": 11328
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001707051006499927,
+      "loss": 0.0073,
+      "step": 11329
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017068590067627757,
+      "loss": 0.0107,
+      "step": 11330
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017066670035706882,
+      "loss": 0.02,
+      "step": 11331
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017064749969268722,
+      "loss": 0.0142,
+      "step": 11332
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017062829868345333,
+      "loss": 0.0017,
+      "step": 11333
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017060909732968796,
+      "loss": 0.0061,
+      "step": 11334
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017058989563171174,
+      "loss": 0.0102,
+      "step": 11335
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001705706935898454,
+      "loss": 0.0012,
+      "step": 11336
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017055149120440968,
+      "loss": 0.0071,
+      "step": 11337
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017053228847572525,
+      "loss": 0.0191,
+      "step": 11338
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001705130854041128,
+      "loss": 0.0179,
+      "step": 11339
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017049388198989314,
+      "loss": 0.0223,
+      "step": 11340
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001704746782333869,
+      "loss": 0.0092,
+      "step": 11341
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000170455474134915,
+      "loss": 0.0038,
+      "step": 11342
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000170436269694798,
+      "loss": 0.0073,
+      "step": 11343
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017041706491335672,
+      "loss": 0.0089,
+      "step": 11344
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017039785979091192,
+      "loss": 0.0109,
+      "step": 11345
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017037865432778442,
+      "loss": 0.009,
+      "step": 11346
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017035944852429483,
+      "loss": 0.0219,
+      "step": 11347
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017034024238076406,
+      "loss": 0.0097,
+      "step": 11348
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017032103589751285,
+      "loss": 0.0124,
+      "step": 11349
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000170301829074862,
+      "loss": 0.0066,
+      "step": 11350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017028262191313222,
+      "loss": 0.0123,
+      "step": 11351
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001702634144126444,
+      "loss": 0.0114,
+      "step": 11352
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017024420657371934,
+      "loss": 0.0086,
+      "step": 11353
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001702249983966778,
+      "loss": 0.0078,
+      "step": 11354
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017020578988184059,
+      "loss": 0.0079,
+      "step": 11355
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001701865810295286,
+      "loss": 0.0073,
+      "step": 11356
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017016737184006255,
+      "loss": 0.0113,
+      "step": 11357
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017014816231376333,
+      "loss": 0.0035,
+      "step": 11358
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017012895245095176,
+      "loss": 0.0038,
+      "step": 11359
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001701097422519487,
+      "loss": 0.0068,
+      "step": 11360
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017009053171707498,
+      "loss": 0.0048,
+      "step": 11361
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017007132084665146,
+      "loss": 0.021,
+      "step": 11362
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000170052109640999,
+      "loss": 0.0068,
+      "step": 11363
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017003289810043845,
+      "loss": 0.0063,
+      "step": 11364
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001700136862252907,
+      "loss": 0.0048,
+      "step": 11365
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001699944740158766,
+      "loss": 0.006,
+      "step": 11366
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016997526147251706,
+      "loss": 0.0126,
+      "step": 11367
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016995604859553292,
+      "loss": 0.0077,
+      "step": 11368
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016993683538524514,
+      "loss": 0.0049,
+      "step": 11369
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016991762184197454,
+      "loss": 0.0158,
+      "step": 11370
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016989840796604203,
+      "loss": 0.0008,
+      "step": 11371
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016987919375776854,
+      "loss": 0.0094,
+      "step": 11372
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016985997921747505,
+      "loss": 0.0121,
+      "step": 11373
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016984076434548243,
+      "loss": 0.0022,
+      "step": 11374
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016982154914211157,
+      "loss": 0.0046,
+      "step": 11375
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001698023336076834,
+      "loss": 0.0187,
+      "step": 11376
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001697831177425189,
+      "loss": 0.0107,
+      "step": 11377
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016976390154693896,
+      "loss": 0.0077,
+      "step": 11378
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001697446850212646,
+      "loss": 0.0198,
+      "step": 11379
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016972546816581673,
+      "loss": 0.0016,
+      "step": 11380
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016970625098091627,
+      "loss": 0.0029,
+      "step": 11381
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016968703346688423,
+      "loss": 0.0008,
+      "step": 11382
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001696678156240416,
+      "loss": 0.0082,
+      "step": 11383
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016964859745270927,
+      "loss": 0.0173,
+      "step": 11384
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016962937895320833,
+      "loss": 0.0079,
+      "step": 11385
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016961016012585967,
+      "loss": 0.0029,
+      "step": 11386
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016959094097098433,
+      "loss": 0.0024,
+      "step": 11387
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016957172148890332,
+      "loss": 0.0093,
+      "step": 11388
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016955250167993762,
+      "loss": 0.0099,
+      "step": 11389
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016953328154440818,
+      "loss": 0.003,
+      "step": 11390
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016951406108263608,
+      "loss": 0.0063,
+      "step": 11391
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016949484029494237,
+      "loss": 0.0055,
+      "step": 11392
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016947561918164797,
+      "loss": 0.0029,
+      "step": 11393
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000169456397743074,
+      "loss": 0.0059,
+      "step": 11394
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016943717597954142,
+      "loss": 0.0127,
+      "step": 11395
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016941795389137135,
+      "loss": 0.0211,
+      "step": 11396
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016939873147888484,
+      "loss": 0.0346,
+      "step": 11397
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016937950874240285,
+      "loss": 0.01,
+      "step": 11398
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001693602856822465,
+      "loss": 0.0025,
+      "step": 11399
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016934106229873677,
+      "loss": 0.0052,
+      "step": 11400
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016932183859219484,
+      "loss": 0.0026,
+      "step": 11401
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016930261456294174,
+      "loss": 0.0023,
+      "step": 11402
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016928339021129852,
+      "loss": 0.0053,
+      "step": 11403
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016926416553758627,
+      "loss": 0.0115,
+      "step": 11404
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001692449405421261,
+      "loss": 0.0024,
+      "step": 11405
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016922571522523918,
+      "loss": 0.004,
+      "step": 11406
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016920648958724643,
+      "loss": 0.0147,
+      "step": 11407
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016918726362846908,
+      "loss": 0.0118,
+      "step": 11408
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001691680373492282,
+      "loss": 0.0043,
+      "step": 11409
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016914881074984488,
+      "loss": 0.0107,
+      "step": 11410
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016912958383064037,
+      "loss": 0.0056,
+      "step": 11411
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001691103565919356,
+      "loss": 0.007,
+      "step": 11412
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016909112903405182,
+      "loss": 0.0211,
+      "step": 11413
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001690719011573102,
+      "loss": 0.0081,
+      "step": 11414
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001690526729620318,
+      "loss": 0.0038,
+      "step": 11415
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001690334444485378,
+      "loss": 0.0134,
+      "step": 11416
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016901421561714934,
+      "loss": 0.0103,
+      "step": 11417
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001689949864681876,
+      "loss": 0.0043,
+      "step": 11418
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016897575700197372,
+      "loss": 0.0065,
+      "step": 11419
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016895652721882892,
+      "loss": 0.0153,
+      "step": 11420
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001689372971190743,
+      "loss": 0.0146,
+      "step": 11421
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016891806670303106,
+      "loss": 0.0072,
+      "step": 11422
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001688988359710204,
+      "loss": 0.0081,
+      "step": 11423
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016887960492336347,
+      "loss": 0.028,
+      "step": 11424
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016886037356038155,
+      "loss": 0.0276,
+      "step": 11425
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001688411418823958,
+      "loss": 0.0035,
+      "step": 11426
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016882190988972738,
+      "loss": 0.0179,
+      "step": 11427
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016880267758269754,
+      "loss": 0.0147,
+      "step": 11428
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001687834449616275,
+      "loss": 0.0057,
+      "step": 11429
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016876421202683854,
+      "loss": 0.0022,
+      "step": 11430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016874497877865174,
+      "loss": 0.001,
+      "step": 11431
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016872574521738842,
+      "loss": 0.005,
+      "step": 11432
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016870651134336982,
+      "loss": 0.0024,
+      "step": 11433
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016868727715691712,
+      "loss": 0.0206,
+      "step": 11434
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016866804265835167,
+      "loss": 0.0072,
+      "step": 11435
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016864880784799464,
+      "loss": 0.0039,
+      "step": 11436
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016862957272616736,
+      "loss": 0.0091,
+      "step": 11437
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016861033729319102,
+      "loss": 0.0055,
+      "step": 11438
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016859110154938695,
+      "loss": 0.0046,
+      "step": 11439
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016857186549507636,
+      "loss": 0.0064,
+      "step": 11440
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016855262913058054,
+      "loss": 0.0025,
+      "step": 11441
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016853339245622082,
+      "loss": 0.0074,
+      "step": 11442
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016851415547231846,
+      "loss": 0.0166,
+      "step": 11443
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016849491817919475,
+      "loss": 0.0128,
+      "step": 11444
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000168475680577171,
+      "loss": 0.0193,
+      "step": 11445
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016845644266656848,
+      "loss": 0.0057,
+      "step": 11446
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016843720444770856,
+      "loss": 0.02,
+      "step": 11447
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016841796592091256,
+      "loss": 0.0162,
+      "step": 11448
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001683987270865017,
+      "loss": 0.0029,
+      "step": 11449
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016837948794479738,
+      "loss": 0.0225,
+      "step": 11450
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016836024849612095,
+      "loss": 0.0051,
+      "step": 11451
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016834100874079367,
+      "loss": 0.0066,
+      "step": 11452
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016832176867913697,
+      "loss": 0.0073,
+      "step": 11453
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016830252831147207,
+      "loss": 0.003,
+      "step": 11454
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016828328763812044,
+      "loss": 0.0358,
+      "step": 11455
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016826404665940342,
+      "loss": 0.0106,
+      "step": 11456
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016824480537564235,
+      "loss": 0.0098,
+      "step": 11457
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016822556378715857,
+      "loss": 0.0094,
+      "step": 11458
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016820632189427345,
+      "loss": 0.0079,
+      "step": 11459
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001681870796973084,
+      "loss": 0.0106,
+      "step": 11460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001681678371965848,
+      "loss": 0.0059,
+      "step": 11461
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016814859439242405,
+      "loss": 0.0012,
+      "step": 11462
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016812935128514747,
+      "loss": 0.0059,
+      "step": 11463
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016811010787507656,
+      "loss": 0.001,
+      "step": 11464
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016809086416253255,
+      "loss": 0.0141,
+      "step": 11465
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016807162014783705,
+      "loss": 0.009,
+      "step": 11466
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016805237583131138,
+      "loss": 0.0047,
+      "step": 11467
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016803313121327697,
+      "loss": 0.0244,
+      "step": 11468
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016801388629405516,
+      "loss": 0.0022,
+      "step": 11469
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016799464107396752,
+      "loss": 0.0093,
+      "step": 11470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001679753955533354,
+      "loss": 0.0004,
+      "step": 11471
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001679561497324802,
+      "loss": 0.0061,
+      "step": 11472
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016793690361172338,
+      "loss": 0.0227,
+      "step": 11473
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016791765719138647,
+      "loss": 0.0255,
+      "step": 11474
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016789841047179082,
+      "loss": 0.0017,
+      "step": 11475
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016787916345325796,
+      "loss": 0.0054,
+      "step": 11476
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016785991613610932,
+      "loss": 0.017,
+      "step": 11477
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016784066852066633,
+      "loss": 0.0056,
+      "step": 11478
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016782142060725054,
+      "loss": 0.0083,
+      "step": 11479
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001678021723961834,
+      "loss": 0.0132,
+      "step": 11480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016778292388778634,
+      "loss": 0.0026,
+      "step": 11481
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016776367508238092,
+      "loss": 0.0032,
+      "step": 11482
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016774442598028856,
+      "loss": 0.0073,
+      "step": 11483
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016772517658183078,
+      "loss": 0.0019,
+      "step": 11484
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016770592688732913,
+      "loss": 0.0038,
+      "step": 11485
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016768667689710505,
+      "loss": 0.0087,
+      "step": 11486
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001676674266114801,
+      "loss": 0.0253,
+      "step": 11487
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016764817603077576,
+      "loss": 0.0065,
+      "step": 11488
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001676289251553136,
+      "loss": 0.0125,
+      "step": 11489
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016760967398541507,
+      "loss": 0.0042,
+      "step": 11490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016759042252140178,
+      "loss": 0.0043,
+      "step": 11491
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016757117076359523,
+      "loss": 0.0055,
+      "step": 11492
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001675519187123169,
+      "loss": 0.0108,
+      "step": 11493
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001675326663678885,
+      "loss": 0.0039,
+      "step": 11494
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016751341373063137,
+      "loss": 0.0031,
+      "step": 11495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016749416080086722,
+      "loss": 0.023,
+      "step": 11496
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016747490757891754,
+      "loss": 0.007,
+      "step": 11497
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016745565406510395,
+      "loss": 0.0052,
+      "step": 11498
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016743640025974798,
+      "loss": 0.0169,
+      "step": 11499
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016741714616317122,
+      "loss": 0.0044,
+      "step": 11500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001673978917756952,
+      "loss": 0.0073,
+      "step": 11501
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001673786370976416,
+      "loss": 0.0042,
+      "step": 11502
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001673593821293319,
+      "loss": 0.0095,
+      "step": 11503
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001673401268710878,
+      "loss": 0.005,
+      "step": 11504
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016732087132323084,
+      "loss": 0.0097,
+      "step": 11505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001673016154860826,
+      "loss": 0.0153,
+      "step": 11506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016728235935996473,
+      "loss": 0.0134,
+      "step": 11507
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001672631029451989,
+      "loss": 0.012,
+      "step": 11508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001672438462421066,
+      "loss": 0.0164,
+      "step": 11509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016722458925100953,
+      "loss": 0.0052,
+      "step": 11510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001672053319722293,
+      "loss": 0.0048,
+      "step": 11511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016718607440608757,
+      "loss": 0.0115,
+      "step": 11512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016716681655290597,
+      "loss": 0.006,
+      "step": 11513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016714755841300608,
+      "loss": 0.0094,
+      "step": 11514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016712829998670963,
+      "loss": 0.007,
+      "step": 11515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001671090412743382,
+      "loss": 0.0171,
+      "step": 11516
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016708978227621355,
+      "loss": 0.0146,
+      "step": 11517
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016707052299265722,
+      "loss": 0.0103,
+      "step": 11518
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016705126342399094,
+      "loss": 0.0044,
+      "step": 11519
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001670320035705364,
+      "loss": 0.0164,
+      "step": 11520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016701274343261524,
+      "loss": 0.0123,
+      "step": 11521
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016699348301054916,
+      "loss": 0.0143,
+      "step": 11522
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016697422230465983,
+      "loss": 0.0213,
+      "step": 11523
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016695496131526893,
+      "loss": 0.0056,
+      "step": 11524
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016693570004269815,
+      "loss": 0.0108,
+      "step": 11525
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016691643848726925,
+      "loss": 0.0031,
+      "step": 11526
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016689717664930388,
+      "loss": 0.013,
+      "step": 11527
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001668779145291238,
+      "loss": 0.0161,
+      "step": 11528
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016685865212705064,
+      "loss": 0.0252,
+      "step": 11529
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001668393894434062,
+      "loss": 0.0064,
+      "step": 11530
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016682012647851216,
+      "loss": 0.0192,
+      "step": 11531
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016680086323269026,
+      "loss": 0.001,
+      "step": 11532
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016678159970626223,
+      "loss": 0.0025,
+      "step": 11533
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016676233589954978,
+      "loss": 0.0081,
+      "step": 11534
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001667430718128747,
+      "loss": 0.0199,
+      "step": 11535
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016672380744655875,
+      "loss": 0.003,
+      "step": 11536
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001667045428009236,
+      "loss": 0.0019,
+      "step": 11537
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016668527787629106,
+      "loss": 0.0078,
+      "step": 11538
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016666601267298285,
+      "loss": 0.0121,
+      "step": 11539
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016664674719132084,
+      "loss": 0.012,
+      "step": 11540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001666274814316267,
+      "loss": 0.008,
+      "step": 11541
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016660821539422223,
+      "loss": 0.009,
+      "step": 11542
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016658894907942923,
+      "loss": 0.0144,
+      "step": 11543
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016656968248756945,
+      "loss": 0.004,
+      "step": 11544
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001665504156189647,
+      "loss": 0.0049,
+      "step": 11545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001665311484739368,
+      "loss": 0.0063,
+      "step": 11546
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016651188105280744,
+      "loss": 0.0201,
+      "step": 11547
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016649261335589856,
+      "loss": 0.0117,
+      "step": 11548
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016647334538353182,
+      "loss": 0.0036,
+      "step": 11549
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016645407713602924,
+      "loss": 0.0202,
+      "step": 11550
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016643480861371242,
+      "loss": 0.0052,
+      "step": 11551
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001664155398169033,
+      "loss": 0.0083,
+      "step": 11552
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001663962707459237,
+      "loss": 0.0022,
+      "step": 11553
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016637700140109544,
+      "loss": 0.0083,
+      "step": 11554
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001663577317827403,
+      "loss": 0.002,
+      "step": 11555
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016633846189118018,
+      "loss": 0.011,
+      "step": 11556
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001663191917267369,
+      "loss": 0.0084,
+      "step": 11557
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016629992128973232,
+      "loss": 0.0041,
+      "step": 11558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016628065058048829,
+      "loss": 0.0084,
+      "step": 11559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016626137959932665,
+      "loss": 0.0102,
+      "step": 11560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016624210834656925,
+      "loss": 0.011,
+      "step": 11561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016622283682253802,
+      "loss": 0.0064,
+      "step": 11562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016620356502755477,
+      "loss": 0.0096,
+      "step": 11563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016618429296194146,
+      "loss": 0.0065,
+      "step": 11564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016616502062601987,
+      "loss": 0.0034,
+      "step": 11565
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001661457480201119,
+      "loss": 0.0195,
+      "step": 11566
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016612647514453946,
+      "loss": 0.0026,
+      "step": 11567
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016610720199962444,
+      "loss": 0.0148,
+      "step": 11568
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016608792858568877,
+      "loss": 0.0044,
+      "step": 11569
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001660686549030543,
+      "loss": 0.0071,
+      "step": 11570
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016604938095204298,
+      "loss": 0.015,
+      "step": 11571
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016603010673297669,
+      "loss": 0.0143,
+      "step": 11572
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001660108322461774,
+      "loss": 0.0082,
+      "step": 11573
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016599155749196694,
+      "loss": 0.0088,
+      "step": 11574
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001659722824706673,
+      "loss": 0.0044,
+      "step": 11575
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001659530071826004,
+      "loss": 0.0351,
+      "step": 11576
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016593373162808814,
+      "loss": 0.0089,
+      "step": 11577
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016591445580745254,
+      "loss": 0.0154,
+      "step": 11578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001658951797210154,
+      "loss": 0.0201,
+      "step": 11579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001658759033690988,
+      "loss": 0.0054,
+      "step": 11580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016585662675202466,
+      "loss": 0.0059,
+      "step": 11581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016583734987011493,
+      "loss": 0.0043,
+      "step": 11582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016581807272369152,
+      "loss": 0.0106,
+      "step": 11583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001657987953130765,
+      "loss": 0.0019,
+      "step": 11584
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016577951763859173,
+      "loss": 0.0046,
+      "step": 11585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016576023970055922,
+      "loss": 0.0073,
+      "step": 11586
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000165740961499301,
+      "loss": 0.0115,
+      "step": 11587
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016572168303513897,
+      "loss": 0.0055,
+      "step": 11588
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016570240430839516,
+      "loss": 0.0073,
+      "step": 11589
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016568312531939155,
+      "loss": 0.0038,
+      "step": 11590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016566384606845017,
+      "loss": 0.0031,
+      "step": 11591
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000165644566555893,
+      "loss": 0.0211,
+      "step": 11592
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016562528678204202,
+      "loss": 0.0132,
+      "step": 11593
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016560600674721927,
+      "loss": 0.0061,
+      "step": 11594
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001655867264517467,
+      "loss": 0.0038,
+      "step": 11595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016556744589594645,
+      "loss": 0.0043,
+      "step": 11596
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016554816508014043,
+      "loss": 0.008,
+      "step": 11597
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001655288840046507,
+      "loss": 0.013,
+      "step": 11598
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001655096026697993,
+      "loss": 0.0045,
+      "step": 11599
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016549032107590823,
+      "loss": 0.0092,
+      "step": 11600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001654710392232996,
+      "loss": 0.0087,
+      "step": 11601
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001654517571122954,
+      "loss": 0.0043,
+      "step": 11602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016543247474321766,
+      "loss": 0.0071,
+      "step": 11603
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001654131921163885,
+      "loss": 0.0027,
+      "step": 11604
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016539390923212997,
+      "loss": 0.0111,
+      "step": 11605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000165374626090764,
+      "loss": 0.0054,
+      "step": 11606
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016535534269261284,
+      "loss": 0.0091,
+      "step": 11607
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016533605903799843,
+      "loss": 0.0137,
+      "step": 11608
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001653167751272429,
+      "loss": 0.0136,
+      "step": 11609
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016529749096066827,
+      "loss": 0.0104,
+      "step": 11610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001652782065385967,
+      "loss": 0.0067,
+      "step": 11611
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016525892186135025,
+      "loss": 0.005,
+      "step": 11612
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016523963692925098,
+      "loss": 0.0051,
+      "step": 11613
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000165220351742621,
+      "loss": 0.0031,
+      "step": 11614
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016520106630178247,
+      "loss": 0.0108,
+      "step": 11615
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001651817806070574,
+      "loss": 0.0077,
+      "step": 11616
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016516249465876794,
+      "loss": 0.0343,
+      "step": 11617
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001651432084572362,
+      "loss": 0.0008,
+      "step": 11618
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016512392200278433,
+      "loss": 0.0205,
+      "step": 11619
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016510463529573432,
+      "loss": 0.0056,
+      "step": 11620
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016508534833640845,
+      "loss": 0.0217,
+      "step": 11621
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016506606112512876,
+      "loss": 0.006,
+      "step": 11622
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016504677366221743,
+      "loss": 0.0119,
+      "step": 11623
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016502748594799662,
+      "loss": 0.0011,
+      "step": 11624
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016500819798278838,
+      "loss": 0.0099,
+      "step": 11625
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016498890976691488,
+      "loss": 0.0169,
+      "step": 11626
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016496962130069836,
+      "loss": 0.0129,
+      "step": 11627
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016495033258446086,
+      "loss": 0.0013,
+      "step": 11628
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016493104361852461,
+      "loss": 0.0143,
+      "step": 11629
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016491175440321173,
+      "loss": 0.0073,
+      "step": 11630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001648924649388444,
+      "loss": 0.0097,
+      "step": 11631
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016487317522574475,
+      "loss": 0.0107,
+      "step": 11632
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001648538852642351,
+      "loss": 0.0036,
+      "step": 11633
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016483459505463747,
+      "loss": 0.0124,
+      "step": 11634
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001648153045972741,
+      "loss": 0.0068,
+      "step": 11635
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016479601389246716,
+      "loss": 0.0052,
+      "step": 11636
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001647767229405389,
+      "loss": 0.0055,
+      "step": 11637
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016475743174181149,
+      "loss": 0.0054,
+      "step": 11638
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001647381402966071,
+      "loss": 0.0082,
+      "step": 11639
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001647188486052479,
+      "loss": 0.0039,
+      "step": 11640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016469955666805614,
+      "loss": 0.0104,
+      "step": 11641
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016468026448535408,
+      "loss": 0.0102,
+      "step": 11642
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016466097205746386,
+      "loss": 0.0141,
+      "step": 11643
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016464167938470776,
+      "loss": 0.0107,
+      "step": 11644
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016462238646740797,
+      "loss": 0.0028,
+      "step": 11645
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001646030933058867,
+      "loss": 0.0154,
+      "step": 11646
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016458379990046625,
+      "loss": 0.0071,
+      "step": 11647
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016456450625146882,
+      "loss": 0.0099,
+      "step": 11648
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001645452123592166,
+      "loss": 0.0077,
+      "step": 11649
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016452591822403189,
+      "loss": 0.0014,
+      "step": 11650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016450662384623693,
+      "loss": 0.0132,
+      "step": 11651
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016448732922615393,
+      "loss": 0.0092,
+      "step": 11652
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016446803436410523,
+      "loss": 0.0044,
+      "step": 11653
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016444873926041304,
+      "loss": 0.0045,
+      "step": 11654
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016442944391539962,
+      "loss": 0.0039,
+      "step": 11655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001644101483293873,
+      "loss": 0.0052,
+      "step": 11656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016439085250269823,
+      "loss": 0.0087,
+      "step": 11657
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016437155643565478,
+      "loss": 0.0048,
+      "step": 11658
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016435226012857923,
+      "loss": 0.0088,
+      "step": 11659
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001643329635817938,
+      "loss": 0.003,
+      "step": 11660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001643136667956209,
+      "loss": 0.0084,
+      "step": 11661
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016429436977038268,
+      "loss": 0.0027,
+      "step": 11662
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016427507250640149,
+      "loss": 0.0018,
+      "step": 11663
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001642557750039997,
+      "loss": 0.0191,
+      "step": 11664
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016423647726349953,
+      "loss": 0.0059,
+      "step": 11665
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016421717928522334,
+      "loss": 0.0104,
+      "step": 11666
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016419788106949336,
+      "loss": 0.0018,
+      "step": 11667
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000164178582616632,
+      "loss": 0.0057,
+      "step": 11668
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016415928392696157,
+      "loss": 0.012,
+      "step": 11669
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016413998500080439,
+      "loss": 0.0107,
+      "step": 11670
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001641206858384827,
+      "loss": 0.0101,
+      "step": 11671
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016410138644031895,
+      "loss": 0.0066,
+      "step": 11672
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016408208680663536,
+      "loss": 0.0033,
+      "step": 11673
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016406278693775438,
+      "loss": 0.0041,
+      "step": 11674
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016404348683399838,
+      "loss": 0.0148,
+      "step": 11675
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016402418649568955,
+      "loss": 0.0049,
+      "step": 11676
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001640048859231504,
+      "loss": 0.0045,
+      "step": 11677
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016398558511670315,
+      "loss": 0.0092,
+      "step": 11678
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016396628407667026,
+      "loss": 0.0121,
+      "step": 11679
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001639469828033741,
+      "loss": 0.018,
+      "step": 11680
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016392768129713694,
+      "loss": 0.0047,
+      "step": 11681
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016390837955828125,
+      "loss": 0.0115,
+      "step": 11682
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016388907758712934,
+      "loss": 0.0076,
+      "step": 11683
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001638697753840036,
+      "loss": 0.0064,
+      "step": 11684
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016385047294922652,
+      "loss": 0.004,
+      "step": 11685
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001638311702831203,
+      "loss": 0.0052,
+      "step": 11686
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016381186738600746,
+      "loss": 0.0008,
+      "step": 11687
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016379256425821037,
+      "loss": 0.0008,
+      "step": 11688
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016377326090005146,
+      "loss": 0.0147,
+      "step": 11689
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016375395731185304,
+      "loss": 0.0074,
+      "step": 11690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001637346534939376,
+      "loss": 0.0055,
+      "step": 11691
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016371534944662752,
+      "loss": 0.0138,
+      "step": 11692
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016369604517024516,
+      "loss": 0.0104,
+      "step": 11693
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016367674066511303,
+      "loss": 0.0021,
+      "step": 11694
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016365743593155352,
+      "loss": 0.0017,
+      "step": 11695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016363813096988904,
+      "loss": 0.0166,
+      "step": 11696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016361882578044202,
+      "loss": 0.0037,
+      "step": 11697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016359952036353498,
+      "loss": 0.0161,
+      "step": 11698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635802147194902,
+      "loss": 0.0145,
+      "step": 11699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635609088486302,
+      "loss": 0.0076,
+      "step": 11700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016354160275127745,
+      "loss": 0.0053,
+      "step": 11701
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635222964277544,
+      "loss": 0.0067,
+      "step": 11702
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635029898783834,
+      "loss": 0.0105,
+      "step": 11703
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016348368310348702,
+      "loss": 0.0009,
+      "step": 11704
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016346437610338766,
+      "loss": 0.0282,
+      "step": 11705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016344506887840785,
+      "loss": 0.0045,
+      "step": 11706
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016342576142887,
+      "loss": 0.0046,
+      "step": 11707
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016340645375509657,
+      "loss": 0.0094,
+      "step": 11708
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016338714585741005,
+      "loss": 0.0268,
+      "step": 11709
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016336783773613291,
+      "loss": 0.0059,
+      "step": 11710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001633485293915877,
+      "loss": 0.0044,
+      "step": 11711
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016332922082409686,
+      "loss": 0.0114,
+      "step": 11712
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001633099120339828,
+      "loss": 0.0059,
+      "step": 11713
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016329060302156814,
+      "loss": 0.0064,
+      "step": 11714
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016327129378717524,
+      "loss": 0.0056,
+      "step": 11715
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016325198433112678,
+      "loss": 0.004,
+      "step": 11716
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001632326746537451,
+      "loss": 0.0085,
+      "step": 11717
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016321336475535285,
+      "loss": 0.0187,
+      "step": 11718
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001631940546362724,
+      "loss": 0.0099,
+      "step": 11719
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016317474429682633,
+      "loss": 0.0177,
+      "step": 11720
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001631554337373372,
+      "loss": 0.0156,
+      "step": 11721
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001631361229581275,
+      "loss": 0.017,
+      "step": 11722
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016311681195951974,
+      "loss": 0.0083,
+      "step": 11723
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001630975007418364,
+      "loss": 0.0051,
+      "step": 11724
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016307818930540014,
+      "loss": 0.0125,
+      "step": 11725
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001630588776505334,
+      "loss": 0.0079,
+      "step": 11726
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016303956577755878,
+      "loss": 0.0062,
+      "step": 11727
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016302025368679878,
+      "loss": 0.0028,
+      "step": 11728
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016300094137857598,
+      "loss": 0.0269,
+      "step": 11729
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016298162885321294,
+      "loss": 0.0104,
+      "step": 11730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016296231611103216,
+      "loss": 0.0225,
+      "step": 11731
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016294300315235623,
+      "loss": 0.0019,
+      "step": 11732
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001629236899775077,
+      "loss": 0.0067,
+      "step": 11733
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001629043765868092,
+      "loss": 0.0133,
+      "step": 11734
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016288506298058322,
+      "loss": 0.0175,
+      "step": 11735
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016286574915915235,
+      "loss": 0.0045,
+      "step": 11736
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001628464351228392,
+      "loss": 0.0191,
+      "step": 11737
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016282712087196634,
+      "loss": 0.0057,
+      "step": 11738
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016280780640685639,
+      "loss": 0.0093,
+      "step": 11739
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016278849172783185,
+      "loss": 0.0185,
+      "step": 11740
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016276917683521533,
+      "loss": 0.0092,
+      "step": 11741
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001627498617293295,
+      "loss": 0.0155,
+      "step": 11742
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001627305464104969,
+      "loss": 0.009,
+      "step": 11743
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016271123087904018,
+      "loss": 0.0041,
+      "step": 11744
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001626919151352818,
+      "loss": 0.0074,
+      "step": 11745
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001626725991795446,
+      "loss": 0.0084,
+      "step": 11746
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000162653283012151,
+      "loss": 0.003,
+      "step": 11747
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001626339666334237,
+      "loss": 0.0063,
+      "step": 11748
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016261465004368533,
+      "loss": 0.0019,
+      "step": 11749
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016259533324325848,
+      "loss": 0.014,
+      "step": 11750
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001625760162324658,
+      "loss": 0.0064,
+      "step": 11751
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016255669901162992,
+      "loss": 0.0076,
+      "step": 11752
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016253738158107347,
+      "loss": 0.0013,
+      "step": 11753
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016251806394111906,
+      "loss": 0.007,
+      "step": 11754
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016249874609208933,
+      "loss": 0.0055,
+      "step": 11755
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016247942803430693,
+      "loss": 0.0041,
+      "step": 11756
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016246010976809458,
+      "loss": 0.0142,
+      "step": 11757
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016244079129377489,
+      "loss": 0.0104,
+      "step": 11758
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016242147261167044,
+      "loss": 0.0244,
+      "step": 11759
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016240215372210402,
+      "loss": 0.0122,
+      "step": 11760
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016238283462539815,
+      "loss": 0.0058,
+      "step": 11761
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016236351532187562,
+      "loss": 0.0174,
+      "step": 11762
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016234419581185905,
+      "loss": 0.0063,
+      "step": 11763
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001623248760956711,
+      "loss": 0.0091,
+      "step": 11764
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016230555617363445,
+      "loss": 0.0008,
+      "step": 11765
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016228623604607175,
+      "loss": 0.0066,
+      "step": 11766
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016226691571330577,
+      "loss": 0.0032,
+      "step": 11767
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016224759517565908,
+      "loss": 0.0049,
+      "step": 11768
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016222827443345444,
+      "loss": 0.0034,
+      "step": 11769
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016220895348701458,
+      "loss": 0.0047,
+      "step": 11770
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016218963233666215,
+      "loss": 0.0052,
+      "step": 11771
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016217031098271986,
+      "loss": 0.0227,
+      "step": 11772
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016215098942551037,
+      "loss": 0.0029,
+      "step": 11773
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001621316676653564,
+      "loss": 0.0152,
+      "step": 11774
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001621123457025807,
+      "loss": 0.0121,
+      "step": 11775
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000162093023537506,
+      "loss": 0.0047,
+      "step": 11776
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016207370117045491,
+      "loss": 0.0326,
+      "step": 11777
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016205437860175026,
+      "loss": 0.0054,
+      "step": 11778
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001620350558317147,
+      "loss": 0.005,
+      "step": 11779
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016201573286067105,
+      "loss": 0.0027,
+      "step": 11780
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016199640968894198,
+      "loss": 0.0043,
+      "step": 11781
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016197708631685018,
+      "loss": 0.0063,
+      "step": 11782
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016195776274471844,
+      "loss": 0.0169,
+      "step": 11783
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016193843897286948,
+      "loss": 0.006,
+      "step": 11784
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001619191150016261,
+      "loss": 0.0115,
+      "step": 11785
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016189979083131096,
+      "loss": 0.0255,
+      "step": 11786
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016188046646224683,
+      "loss": 0.0116,
+      "step": 11787
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016186114189475648,
+      "loss": 0.0156,
+      "step": 11788
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016184181712916273,
+      "loss": 0.0074,
+      "step": 11789
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001618224921657883,
+      "loss": 0.0011,
+      "step": 11790
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016180316700495587,
+      "loss": 0.0067,
+      "step": 11791
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016178384164698827,
+      "loss": 0.0084,
+      "step": 11792
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001617645160922083,
+      "loss": 0.0171,
+      "step": 11793
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001617451903409387,
+      "loss": 0.0157,
+      "step": 11794
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016172586439350227,
+      "loss": 0.004,
+      "step": 11795
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016170653825022175,
+      "loss": 0.0076,
+      "step": 11796
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016168721191141993,
+      "loss": 0.0042,
+      "step": 11797
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001616678853774196,
+      "loss": 0.0046,
+      "step": 11798
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016164855864854357,
+      "loss": 0.007,
+      "step": 11799
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016162923172511465,
+      "loss": 0.0011,
+      "step": 11800
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016160990460745557,
+      "loss": 0.0028,
+      "step": 11801
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615905772958892,
+      "loss": 0.0086,
+      "step": 11802
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615712497907383,
+      "loss": 0.003,
+      "step": 11803
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615519220923257,
+      "loss": 0.0107,
+      "step": 11804
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615325942009742,
+      "loss": 0.0035,
+      "step": 11805
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016151326611700656,
+      "loss": 0.0091,
+      "step": 11806
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016149393784074568,
+      "loss": 0.0043,
+      "step": 11807
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016147460937251433,
+      "loss": 0.0135,
+      "step": 11808
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016145528071263534,
+      "loss": 0.0142,
+      "step": 11809
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016143595186143153,
+      "loss": 0.0077,
+      "step": 11810
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016141662281922577,
+      "loss": 0.0193,
+      "step": 11811
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001613972935863408,
+      "loss": 0.0025,
+      "step": 11812
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016137796416309958,
+      "loss": 0.0247,
+      "step": 11813
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016135863454982488,
+      "loss": 0.0066,
+      "step": 11814
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016133930474683953,
+      "loss": 0.0092,
+      "step": 11815
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016131997475446634,
+      "loss": 0.0016,
+      "step": 11816
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016130064457302822,
+      "loss": 0.0123,
+      "step": 11817
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016128131420284806,
+      "loss": 0.0161,
+      "step": 11818
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001612619836442486,
+      "loss": 0.0046,
+      "step": 11819
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016124265289755275,
+      "loss": 0.0169,
+      "step": 11820
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001612233219630834,
+      "loss": 0.015,
+      "step": 11821
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016120399084116338,
+      "loss": 0.0063,
+      "step": 11822
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001611846595321156,
+      "loss": 0.0193,
+      "step": 11823
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016116532803626287,
+      "loss": 0.014,
+      "step": 11824
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001611459963539281,
+      "loss": 0.0163,
+      "step": 11825
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001611266644854341,
+      "loss": 0.0083,
+      "step": 11826
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016110733243110384,
+      "loss": 0.0041,
+      "step": 11827
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016108800019126013,
+      "loss": 0.0109,
+      "step": 11828
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001610686677662259,
+      "loss": 0.0034,
+      "step": 11829
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016104933515632402,
+      "loss": 0.0099,
+      "step": 11830
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001610300023618774,
+      "loss": 0.0173,
+      "step": 11831
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016101066938320897,
+      "loss": 0.0036,
+      "step": 11832
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001609913362206415,
+      "loss": 0.0036,
+      "step": 11833
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000160972002874498,
+      "loss": 0.0124,
+      "step": 11834
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016095266934510133,
+      "loss": 0.0052,
+      "step": 11835
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001609333356327744,
+      "loss": 0.0157,
+      "step": 11836
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001609140017378402,
+      "loss": 0.0112,
+      "step": 11837
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016089466766062147,
+      "loss": 0.0136,
+      "step": 11838
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016087533340144127,
+      "loss": 0.015,
+      "step": 11839
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016085599896062242,
+      "loss": 0.0043,
+      "step": 11840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016083666433848798,
+      "loss": 0.0016,
+      "step": 11841
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016081732953536075,
+      "loss": 0.0171,
+      "step": 11842
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016079799455156368,
+      "loss": 0.0302,
+      "step": 11843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016077865938741976,
+      "loss": 0.0179,
+      "step": 11844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016075932404325182,
+      "loss": 0.0217,
+      "step": 11845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016073998851938293,
+      "loss": 0.0055,
+      "step": 11846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016072065281613594,
+      "loss": 0.0107,
+      "step": 11847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001607013169338338,
+      "loss": 0.0091,
+      "step": 11848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016068198087279946,
+      "loss": 0.0034,
+      "step": 11849
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016066264463335587,
+      "loss": 0.0096,
+      "step": 11850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000160643308215826,
+      "loss": 0.0029,
+      "step": 11851
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016062397162053277,
+      "loss": 0.0103,
+      "step": 11852
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016060463484779918,
+      "loss": 0.003,
+      "step": 11853
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016058529789794815,
+      "loss": 0.0084,
+      "step": 11854
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016056596077130273,
+      "loss": 0.0111,
+      "step": 11855
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001605466234681858,
+      "loss": 0.006,
+      "step": 11856
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016052728598892028,
+      "loss": 0.0053,
+      "step": 11857
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016050794833382927,
+      "loss": 0.0181,
+      "step": 11858
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001604886105032357,
+      "loss": 0.006,
+      "step": 11859
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001604692724974625,
+      "loss": 0.008,
+      "step": 11860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016044993431683265,
+      "loss": 0.0057,
+      "step": 11861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016043059596166926,
+      "loss": 0.0092,
+      "step": 11862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016041125743229516,
+      "loss": 0.0052,
+      "step": 11863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016039191872903346,
+      "loss": 0.0081,
+      "step": 11864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016037257985220706,
+      "loss": 0.0023,
+      "step": 11865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016035324080213904,
+      "loss": 0.0061,
+      "step": 11866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001603339015791523,
+      "loss": 0.0184,
+      "step": 11867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016031456218356993,
+      "loss": 0.0034,
+      "step": 11868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016029522261571495,
+      "loss": 0.0126,
+      "step": 11869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001602758828759102,
+      "loss": 0.0143,
+      "step": 11870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016025654296447888,
+      "loss": 0.0031,
+      "step": 11871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001602372028817439,
+      "loss": 0.0064,
+      "step": 11872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016021786262802835,
+      "loss": 0.0168,
+      "step": 11873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016019852220365522,
+      "loss": 0.009,
+      "step": 11874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001601791816089475,
+      "loss": 0.0071,
+      "step": 11875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001601598408442282,
+      "loss": 0.0092,
+      "step": 11876
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001601404999098204,
+      "loss": 0.0068,
+      "step": 11877
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016012115880604716,
+      "loss": 0.0023,
+      "step": 11878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016010181753323138,
+      "loss": 0.0076,
+      "step": 11879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016008247609169623,
+      "loss": 0.0114,
+      "step": 11880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016006313448176465,
+      "loss": 0.0067,
+      "step": 11881
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016004379270375974,
+      "loss": 0.0083,
+      "step": 11882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001600244507580046,
+      "loss": 0.011,
+      "step": 11883
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016000510864482215,
+      "loss": 0.0069,
+      "step": 11884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015998576636453552,
+      "loss": 0.0238,
+      "step": 11885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015996642391746772,
+      "loss": 0.0259,
+      "step": 11886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015994708130394184,
+      "loss": 0.0121,
+      "step": 11887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015992773852428093,
+      "loss": 0.0039,
+      "step": 11888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015990839557880805,
+      "loss": 0.0106,
+      "step": 11889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015988905246784626,
+      "loss": 0.0069,
+      "step": 11890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001598697091917186,
+      "loss": 0.0036,
+      "step": 11891
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001598503657507482,
+      "loss": 0.0017,
+      "step": 11892
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015983102214525807,
+      "loss": 0.0104,
+      "step": 11893
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015981167837557134,
+      "loss": 0.0056,
+      "step": 11894
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015979233444201105,
+      "loss": 0.0053,
+      "step": 11895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015977299034490027,
+      "loss": 0.0133,
+      "step": 11896
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015975364608456215,
+      "loss": 0.0053,
+      "step": 11897
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001597343016613197,
+      "loss": 0.0015,
+      "step": 11898
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015971495707549604,
+      "loss": 0.0108,
+      "step": 11899
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015969561232741426,
+      "loss": 0.019,
+      "step": 11900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015967626741739745,
+      "loss": 0.0051,
+      "step": 11901
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015965692234576868,
+      "loss": 0.0143,
+      "step": 11902
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015963757711285112,
+      "loss": 0.0041,
+      "step": 11903
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001596182317189678,
+      "loss": 0.0056,
+      "step": 11904
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015959888616444187,
+      "loss": 0.0089,
+      "step": 11905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015957954044959648,
+      "loss": 0.0052,
+      "step": 11906
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001595601945747546,
+      "loss": 0.0035,
+      "step": 11907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015954084854023944,
+      "loss": 0.0039,
+      "step": 11908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001595215023463741,
+      "loss": 0.0134,
+      "step": 11909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015950215599348173,
+      "loss": 0.0106,
+      "step": 11910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015948280948188537,
+      "loss": 0.0061,
+      "step": 11911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015946346281190822,
+      "loss": 0.0096,
+      "step": 11912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015944411598387337,
+      "loss": 0.0112,
+      "step": 11913
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015942476899810396,
+      "loss": 0.0042,
+      "step": 11914
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015940542185492314,
+      "loss": 0.0135,
+      "step": 11915
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000159386074554654,
+      "loss": 0.0086,
+      "step": 11916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001593667270976197,
+      "loss": 0.0054,
+      "step": 11917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015934737948414333,
+      "loss": 0.0105,
+      "step": 11918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015932803171454812,
+      "loss": 0.0335,
+      "step": 11919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015930868378915725,
+      "loss": 0.0144,
+      "step": 11920
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001592893357082937,
+      "loss": 0.0025,
+      "step": 11921
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001592699874722807,
+      "loss": 0.0018,
+      "step": 11922
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001592506390814414,
+      "loss": 0.0023,
+      "step": 11923
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000159231290536099,
+      "loss": 0.0086,
+      "step": 11924
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015921194183657664,
+      "loss": 0.0048,
+      "step": 11925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015919259298319743,
+      "loss": 0.0071,
+      "step": 11926
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015917324397628457,
+      "loss": 0.0076,
+      "step": 11927
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015915389481616123,
+      "loss": 0.004,
+      "step": 11928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015913454550315058,
+      "loss": 0.0047,
+      "step": 11929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015911519603757578,
+      "loss": 0.0073,
+      "step": 11930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015909584641975999,
+      "loss": 0.0015,
+      "step": 11931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015907649665002635,
+      "loss": 0.0041,
+      "step": 11932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015905714672869817,
+      "loss": 0.0107,
+      "step": 11933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015903779665609846,
+      "loss": 0.0183,
+      "step": 11934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015901844643255051,
+      "loss": 0.0059,
+      "step": 11935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001589990960583775,
+      "loss": 0.0057,
+      "step": 11936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015897974553390258,
+      "loss": 0.0017,
+      "step": 11937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000158960394859449,
+      "loss": 0.0024,
+      "step": 11938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001589410440353399,
+      "loss": 0.0133,
+      "step": 11939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015892169306189844,
+      "loss": 0.0186,
+      "step": 11940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015890234193944788,
+      "loss": 0.0158,
+      "step": 11941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015888299066831144,
+      "loss": 0.0086,
+      "step": 11942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015886363924881225,
+      "loss": 0.0056,
+      "step": 11943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015884428768127358,
+      "loss": 0.0059,
+      "step": 11944
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015882493596601857,
+      "loss": 0.0018,
+      "step": 11945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015880558410337052,
+      "loss": 0.0183,
+      "step": 11946
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015878623209365255,
+      "loss": 0.016,
+      "step": 11947
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015876687993718796,
+      "loss": 0.0221,
+      "step": 11948
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015874752763429993,
+      "loss": 0.0391,
+      "step": 11949
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015872817518531164,
+      "loss": 0.0026,
+      "step": 11950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015870882259054635,
+      "loss": 0.0049,
+      "step": 11951
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015868946985032734,
+      "loss": 0.0061,
+      "step": 11952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001586701169649777,
+      "loss": 0.0047,
+      "step": 11953
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001586507639348208,
+      "loss": 0.0044,
+      "step": 11954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015863141076017978,
+      "loss": 0.0062,
+      "step": 11955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001586120574413779,
+      "loss": 0.0013,
+      "step": 11956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015859270397873846,
+      "loss": 0.0081,
+      "step": 11957
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001585733503725846,
+      "loss": 0.0027,
+      "step": 11958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015855399662323964,
+      "loss": 0.002,
+      "step": 11959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015853464273102674,
+      "loss": 0.0082,
+      "step": 11960
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001585152886962692,
+      "loss": 0.0058,
+      "step": 11961
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015849593451929033,
+      "loss": 0.0016,
+      "step": 11962
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015847658020041327,
+      "loss": 0.0203,
+      "step": 11963
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584572257399613,
+      "loss": 0.0185,
+      "step": 11964
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584378711382577,
+      "loss": 0.0123,
+      "step": 11965
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015841851639562576,
+      "loss": 0.0074,
+      "step": 11966
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015839916151238868,
+      "loss": 0.0032,
+      "step": 11967
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015837980648886977,
+      "loss": 0.0098,
+      "step": 11968
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015836045132539226,
+      "loss": 0.0103,
+      "step": 11969
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001583410960222794,
+      "loss": 0.007,
+      "step": 11970
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015832174057985455,
+      "loss": 0.0148,
+      "step": 11971
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001583023849984409,
+      "loss": 0.0083,
+      "step": 11972
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015828302927836177,
+      "loss": 0.0124,
+      "step": 11973
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015826367341994038,
+      "loss": 0.0059,
+      "step": 11974
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001582443174235,
+      "loss": 0.0041,
+      "step": 11975
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015822496128936405,
+      "loss": 0.0214,
+      "step": 11976
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015820560501785566,
+      "loss": 0.0123,
+      "step": 11977
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001581862486092982,
+      "loss": 0.0082,
+      "step": 11978
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015816689206401495,
+      "loss": 0.021,
+      "step": 11979
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001581475353823292,
+      "loss": 0.0151,
+      "step": 11980
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015812817856456418,
+      "loss": 0.0119,
+      "step": 11981
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015810882161104324,
+      "loss": 0.0048,
+      "step": 11982
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015808946452208972,
+      "loss": 0.0101,
+      "step": 11983
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015807010729802684,
+      "loss": 0.0057,
+      "step": 11984
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015805074993917792,
+      "loss": 0.0102,
+      "step": 11985
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015803139244586627,
+      "loss": 0.0084,
+      "step": 11986
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015801203481841523,
+      "loss": 0.012,
+      "step": 11987
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015799267705714809,
+      "loss": 0.0015,
+      "step": 11988
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015797331916238817,
+      "loss": 0.0144,
+      "step": 11989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015795396113445876,
+      "loss": 0.0139,
+      "step": 11990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015793460297368315,
+      "loss": 0.0039,
+      "step": 11991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015791524468038474,
+      "loss": 0.0093,
+      "step": 11992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015789588625488677,
+      "loss": 0.0093,
+      "step": 11993
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001578765276975126,
+      "loss": 0.0144,
+      "step": 11994
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015785716900858557,
+      "loss": 0.0007,
+      "step": 11995
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015783781018842896,
+      "loss": 0.0119,
+      "step": 11996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015781845123736613,
+      "loss": 0.0024,
+      "step": 11997
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015779909215572042,
+      "loss": 0.0121,
+      "step": 11998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015777973294381516,
+      "loss": 0.0089,
+      "step": 11999
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015776037360197367,
+      "loss": 0.0004,
+      "step": 12000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015774101413051927,
+      "loss": 0.0141,
+      "step": 12001
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001577216545297753,
+      "loss": 0.0261,
+      "step": 12002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001577022948000652,
+      "loss": 0.0101,
+      "step": 12003
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015768293494171216,
+      "loss": 0.0193,
+      "step": 12004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015766357495503963,
+      "loss": 0.0076,
+      "step": 12005
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015764421484037094,
+      "loss": 0.0211,
+      "step": 12006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015762485459802942,
+      "loss": 0.0061,
+      "step": 12007
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015760549422833846,
+      "loss": 0.0198,
+      "step": 12008
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015758613373162138,
+      "loss": 0.0031,
+      "step": 12009
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015756677310820155,
+      "loss": 0.0062,
+      "step": 12010
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001575474123584023,
+      "loss": 0.0073,
+      "step": 12011
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015752805148254708,
+      "loss": 0.0065,
+      "step": 12012
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001575086904809591,
+      "loss": 0.0046,
+      "step": 12013
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015748932935396188,
+      "loss": 0.0101,
+      "step": 12014
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015746996810187872,
+      "loss": 0.0065,
+      "step": 12015
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015745060672503296,
+      "loss": 0.0209,
+      "step": 12016
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015743124522374802,
+      "loss": 0.0082,
+      "step": 12017
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015741188359834723,
+      "loss": 0.0015,
+      "step": 12018
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015739252184915405,
+      "loss": 0.0064,
+      "step": 12019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015737315997649176,
+      "loss": 0.0202,
+      "step": 12020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015735379798068375,
+      "loss": 0.0094,
+      "step": 12021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015733443586205353,
+      "loss": 0.0119,
+      "step": 12022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001573150736209243,
+      "loss": 0.0088,
+      "step": 12023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001572957112576196,
+      "loss": 0.02,
+      "step": 12024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001572763487724627,
+      "loss": 0.0116,
+      "step": 12025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015725698616577704,
+      "loss": 0.0058,
+      "step": 12026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015723762343788602,
+      "loss": 0.0129,
+      "step": 12027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015721826058911304,
+      "loss": 0.0153,
+      "step": 12028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015719889761978145,
+      "loss": 0.0112,
+      "step": 12029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001571795345302147,
+      "loss": 0.0039,
+      "step": 12030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015716017132073618,
+      "loss": 0.0107,
+      "step": 12031
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015714080799166932,
+      "loss": 0.0217,
+      "step": 12032
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015712144454333742,
+      "loss": 0.0029,
+      "step": 12033
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015710208097606396,
+      "loss": 0.0088,
+      "step": 12034
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015708271729017237,
+      "loss": 0.0125,
+      "step": 12035
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015706335348598604,
+      "loss": 0.0148,
+      "step": 12036
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015704398956382833,
+      "loss": 0.0113,
+      "step": 12037
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015702462552402276,
+      "loss": 0.0064,
+      "step": 12038
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001570052613668926,
+      "loss": 0.0128,
+      "step": 12039
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015698589709276146,
+      "loss": 0.0072,
+      "step": 12040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015696653270195256,
+      "loss": 0.0195,
+      "step": 12041
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015694716819478948,
+      "loss": 0.007,
+      "step": 12042
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015692780357159553,
+      "loss": 0.0092,
+      "step": 12043
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001569084388326942,
+      "loss": 0.0014,
+      "step": 12044
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001568890739784089,
+      "loss": 0.0226,
+      "step": 12045
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001568697090090631,
+      "loss": 0.0082,
+      "step": 12046
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015685034392498012,
+      "loss": 0.0051,
+      "step": 12047
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015683097872648346,
+      "loss": 0.0149,
+      "step": 12048
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001568116134138966,
+      "loss": 0.0111,
+      "step": 12049
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015679224798754296,
+      "loss": 0.0122,
+      "step": 12050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015677288244774593,
+      "loss": 0.0142,
+      "step": 12051
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015675351679482897,
+      "loss": 0.006,
+      "step": 12052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015673415102911552,
+      "loss": 0.0059,
+      "step": 12053
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001567147851509291,
+      "loss": 0.0148,
+      "step": 12054
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015669541916059305,
+      "loss": 0.0148,
+      "step": 12055
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015667605305843085,
+      "loss": 0.0063,
+      "step": 12056
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015665668684476598,
+      "loss": 0.0039,
+      "step": 12057
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015663732051992188,
+      "loss": 0.0069,
+      "step": 12058
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015661795408422195,
+      "loss": 0.0149,
+      "step": 12059
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015659858753798973,
+      "loss": 0.0034,
+      "step": 12060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015657922088154863,
+      "loss": 0.0065,
+      "step": 12061
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015655985411522214,
+      "loss": 0.0021,
+      "step": 12062
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015654048723933375,
+      "loss": 0.0116,
+      "step": 12063
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015652112025420683,
+      "loss": 0.0018,
+      "step": 12064
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015650175316016488,
+      "loss": 0.0162,
+      "step": 12065
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001564823859575314,
+      "loss": 0.014,
+      "step": 12066
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001564630186466298,
+      "loss": 0.0228,
+      "step": 12067
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001564436512277836,
+      "loss": 0.0048,
+      "step": 12068
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015642428370131627,
+      "loss": 0.0124,
+      "step": 12069
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015640491606755129,
+      "loss": 0.0021,
+      "step": 12070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001563855483268121,
+      "loss": 0.0014,
+      "step": 12071
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015636618047942222,
+      "loss": 0.0048,
+      "step": 12072
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015634681252570514,
+      "loss": 0.0088,
+      "step": 12073
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015632744446598425,
+      "loss": 0.0342,
+      "step": 12074
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015630807630058314,
+      "loss": 0.0162,
+      "step": 12075
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001562887080298252,
+      "loss": 0.0076,
+      "step": 12076
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015626933965403402,
+      "loss": 0.0042,
+      "step": 12077
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000156249971173533,
+      "loss": 0.0354,
+      "step": 12078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015623060258864563,
+      "loss": 0.0059,
+      "step": 12079
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001562112338996955,
+      "loss": 0.0068,
+      "step": 12080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015619186510700597,
+      "loss": 0.0004,
+      "step": 12081
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015617249621090068,
+      "loss": 0.015,
+      "step": 12082
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015615312721170303,
+      "loss": 0.0033,
+      "step": 12083
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001561337581097365,
+      "loss": 0.0104,
+      "step": 12084
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001561143889053247,
+      "loss": 0.0082,
+      "step": 12085
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015609501959879104,
+      "loss": 0.0044,
+      "step": 12086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015607565019045902,
+      "loss": 0.009,
+      "step": 12087
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015605628068065218,
+      "loss": 0.0107,
+      "step": 12088
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000156036911069694,
+      "loss": 0.0183,
+      "step": 12089
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000156017541357908,
+      "loss": 0.0265,
+      "step": 12090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001559981715456178,
+      "loss": 0.0096,
+      "step": 12091
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015597880163314672,
+      "loss": 0.0157,
+      "step": 12092
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001559594316208184,
+      "loss": 0.0086,
+      "step": 12093
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001559400615089563,
+      "loss": 0.0071,
+      "step": 12094
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015592069129788397,
+      "loss": 0.0018,
+      "step": 12095
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015590132098792497,
+      "loss": 0.0021,
+      "step": 12096
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015588195057940273,
+      "loss": 0.016,
+      "step": 12097
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015586258007264082,
+      "loss": 0.0041,
+      "step": 12098
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015584320946796274,
+      "loss": 0.0223,
+      "step": 12099
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000155823838765692,
+      "loss": 0.0028,
+      "step": 12100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001558044679661522,
+      "loss": 0.005,
+      "step": 12101
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001557850970696668,
+      "loss": 0.0154,
+      "step": 12102
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001557657260765594,
+      "loss": 0.0065,
+      "step": 12103
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015574635498715348,
+      "loss": 0.0118,
+      "step": 12104
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001557269838017726,
+      "loss": 0.0019,
+      "step": 12105
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015570761252074026,
+      "loss": 0.0067,
+      "step": 12106
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015568824114438,
+      "loss": 0.0067,
+      "step": 12107
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001556688696730154,
+      "loss": 0.0094,
+      "step": 12108
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015564949810696995,
+      "loss": 0.0101,
+      "step": 12109
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015563012644656725,
+      "loss": 0.0058,
+      "step": 12110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001556107546921308,
+      "loss": 0.01,
+      "step": 12111
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015559138284398417,
+      "loss": 0.04,
+      "step": 12112
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015557201090245085,
+      "loss": 0.0106,
+      "step": 12113
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015555263886785447,
+      "loss": 0.0025,
+      "step": 12114
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015553326674051852,
+      "loss": 0.0035,
+      "step": 12115
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001555138945207666,
+      "loss": 0.0074,
+      "step": 12116
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015549452220892218,
+      "loss": 0.0067,
+      "step": 12117
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015547514980530892,
+      "loss": 0.0023,
+      "step": 12118
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015545577731025027,
+      "loss": 0.0163,
+      "step": 12119
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015543640472406986,
+      "loss": 0.0168,
+      "step": 12120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015541703204709125,
+      "loss": 0.0068,
+      "step": 12121
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015539765927963795,
+      "loss": 0.015,
+      "step": 12122
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001553782864220336,
+      "loss": 0.0199,
+      "step": 12123
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015535891347460168,
+      "loss": 0.0146,
+      "step": 12124
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015533954043766578,
+      "loss": 0.0051,
+      "step": 12125
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015532016731154949,
+      "loss": 0.0228,
+      "step": 12126
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015530079409657637,
+      "loss": 0.0089,
+      "step": 12127
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015528142079307,
+      "loss": 0.0038,
+      "step": 12128
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001552620474013539,
+      "loss": 0.0028,
+      "step": 12129
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001552426739217517,
+      "loss": 0.0056,
+      "step": 12130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015522330035458687,
+      "loss": 0.0042,
+      "step": 12131
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015520392670018311,
+      "loss": 0.0103,
+      "step": 12132
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000155184552958864,
+      "loss": 0.0035,
+      "step": 12133
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015516517913095307,
+      "loss": 0.0018,
+      "step": 12134
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015514580521677385,
+      "loss": 0.0039,
+      "step": 12135
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015512643121664999,
+      "loss": 0.0014,
+      "step": 12136
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015510705713090505,
+      "loss": 0.0095,
+      "step": 12137
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015508768295986261,
+      "loss": 0.0172,
+      "step": 12138
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015506830870384627,
+      "loss": 0.0222,
+      "step": 12139
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015504893436317957,
+      "loss": 0.0047,
+      "step": 12140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015502955993818617,
+      "loss": 0.0048,
+      "step": 12141
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015501018542918963,
+      "loss": 0.0047,
+      "step": 12142
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015499081083651352,
+      "loss": 0.0004,
+      "step": 12143
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015497143616048143,
+      "loss": 0.0105,
+      "step": 12144
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000154952061401417,
+      "loss": 0.0111,
+      "step": 12145
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015493268655964379,
+      "loss": 0.0071,
+      "step": 12146
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015491331163548544,
+      "loss": 0.0077,
+      "step": 12147
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001548939366292655,
+      "loss": 0.0076,
+      "step": 12148
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015487456154130753,
+      "loss": 0.0135,
+      "step": 12149
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015485518637193524,
+      "loss": 0.0215,
+      "step": 12150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001548358111214721,
+      "loss": 0.0008,
+      "step": 12151
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015481643579024187,
+      "loss": 0.0046,
+      "step": 12152
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015479706037856798,
+      "loss": 0.0074,
+      "step": 12153
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001547776848867742,
+      "loss": 0.0065,
+      "step": 12154
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015475830931518402,
+      "loss": 0.0038,
+      "step": 12155
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015473893366412114,
+      "loss": 0.0065,
+      "step": 12156
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001547195579339091,
+      "loss": 0.0048,
+      "step": 12157
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015470018212487152,
+      "loss": 0.0005,
+      "step": 12158
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015468080623733203,
+      "loss": 0.0068,
+      "step": 12159
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001546614302716143,
+      "loss": 0.0175,
+      "step": 12160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001546420542280418,
+      "loss": 0.0103,
+      "step": 12161
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015462267810693823,
+      "loss": 0.0157,
+      "step": 12162
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015460330190862723,
+      "loss": 0.013,
+      "step": 12163
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001545839256334324,
+      "loss": 0.0075,
+      "step": 12164
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015456454928167743,
+      "loss": 0.007,
+      "step": 12165
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015454517285368578,
+      "loss": 0.0028,
+      "step": 12166
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001545257963497812,
+      "loss": 0.0217,
+      "step": 12167
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015450641977028727,
+      "loss": 0.0035,
+      "step": 12168
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015448704311552761,
+      "loss": 0.0096,
+      "step": 12169
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015446766638582593,
+      "loss": 0.004,
+      "step": 12170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015444828958150573,
+      "loss": 0.0113,
+      "step": 12171
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015442891270289066,
+      "loss": 0.0106,
+      "step": 12172
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001544095357503044,
+      "loss": 0.0139,
+      "step": 12173
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015439015872407066,
+      "loss": 0.0302,
+      "step": 12174
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001543707816245129,
+      "loss": 0.0024,
+      "step": 12175
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015435140445195484,
+      "loss": 0.0072,
+      "step": 12176
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015433202720672014,
+      "loss": 0.0022,
+      "step": 12177
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001543126498891324,
+      "loss": 0.0099,
+      "step": 12178
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015429327249951528,
+      "loss": 0.0156,
+      "step": 12179
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001542738950381924,
+      "loss": 0.0014,
+      "step": 12180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015425451750548737,
+      "loss": 0.0076,
+      "step": 12181
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001542351399017239,
+      "loss": 0.0182,
+      "step": 12182
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001542157622272256,
+      "loss": 0.0116,
+      "step": 12183
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015419638448231607,
+      "loss": 0.0034,
+      "step": 12184
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015417700666731904,
+      "loss": 0.0026,
+      "step": 12185
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001541576287825581,
+      "loss": 0.0049,
+      "step": 12186
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001541382508283569,
+      "loss": 0.0126,
+      "step": 12187
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015411887280503917,
+      "loss": 0.0058,
+      "step": 12188
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001540994947129284,
+      "loss": 0.0036,
+      "step": 12189
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015408011655234838,
+      "loss": 0.0035,
+      "step": 12190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015406073832362267,
+      "loss": 0.0081,
+      "step": 12191
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000154041360027075,
+      "loss": 0.0046,
+      "step": 12192
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015402198166302898,
+      "loss": 0.0156,
+      "step": 12193
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015400260323180826,
+      "loss": 0.0137,
+      "step": 12194
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001539832247337365,
+      "loss": 0.017,
+      "step": 12195
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015396384616913743,
+      "loss": 0.0057,
+      "step": 12196
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015394446753833461,
+      "loss": 0.0097,
+      "step": 12197
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015392508884165177,
+      "loss": 0.0032,
+      "step": 12198
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001539057100794125,
+      "loss": 0.0091,
+      "step": 12199
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015388633125194048,
+      "loss": 0.0123,
+      "step": 12200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001538669523595594,
+      "loss": 0.0056,
+      "step": 12201
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015384757340259294,
+      "loss": 0.01,
+      "step": 12202
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015382819438136471,
+      "loss": 0.0051,
+      "step": 12203
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015380881529619843,
+      "loss": 0.0012,
+      "step": 12204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015378943614741776,
+      "loss": 0.0108,
+      "step": 12205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015377005693534634,
+      "loss": 0.0063,
+      "step": 12206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015375067766030788,
+      "loss": 0.0201,
+      "step": 12207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015373129832262598,
+      "loss": 0.0198,
+      "step": 12208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001537119189226244,
+      "loss": 0.0011,
+      "step": 12209
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001536925394606267,
+      "loss": 0.0166,
+      "step": 12210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015367315993695667,
+      "loss": 0.0072,
+      "step": 12211
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015365378035193795,
+      "loss": 0.0027,
+      "step": 12212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015363440070589415,
+      "loss": 0.0122,
+      "step": 12213
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015361502099914897,
+      "loss": 0.008,
+      "step": 12214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015359564123202616,
+      "loss": 0.0144,
+      "step": 12215
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001535762614048494,
+      "loss": 0.0087,
+      "step": 12216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015355688151794228,
+      "loss": 0.0007,
+      "step": 12217
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015353750157162853,
+      "loss": 0.0072,
+      "step": 12218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015351812156623183,
+      "loss": 0.0035,
+      "step": 12219
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015349874150207582,
+      "loss": 0.0031,
+      "step": 12220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001534793613794843,
+      "loss": 0.0051,
+      "step": 12221
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015345998119878083,
+      "loss": 0.0063,
+      "step": 12222
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015344060096028915,
+      "loss": 0.0088,
+      "step": 12223
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001534212206643329,
+      "loss": 0.0038,
+      "step": 12224
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015340184031123588,
+      "loss": 0.0084,
+      "step": 12225
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015338245990132165,
+      "loss": 0.0029,
+      "step": 12226
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015336307943491397,
+      "loss": 0.0185,
+      "step": 12227
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015334369891233652,
+      "loss": 0.0108,
+      "step": 12228
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015332431833391298,
+      "loss": 0.019,
+      "step": 12229
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001533049376999671,
+      "loss": 0.0222,
+      "step": 12230
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015328555701082247,
+      "loss": 0.0024,
+      "step": 12231
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015326617626680282,
+      "loss": 0.0009,
+      "step": 12232
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001532467954682319,
+      "loss": 0.0035,
+      "step": 12233
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015322741461543332,
+      "loss": 0.022,
+      "step": 12234
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015320803370873085,
+      "loss": 0.0138,
+      "step": 12235
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015318865274844816,
+      "loss": 0.001,
+      "step": 12236
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015316927173490897,
+      "loss": 0.0078,
+      "step": 12237
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015314989066843695,
+      "loss": 0.0132,
+      "step": 12238
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015313050954935583,
+      "loss": 0.0029,
+      "step": 12239
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015311112837798925,
+      "loss": 0.0207,
+      "step": 12240
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015309174715466095,
+      "loss": 0.0133,
+      "step": 12241
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015307236587969468,
+      "loss": 0.0278,
+      "step": 12242
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015305298455341407,
+      "loss": 0.0079,
+      "step": 12243
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015303360317614283,
+      "loss": 0.0097,
+      "step": 12244
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015301422174820471,
+      "loss": 0.0148,
+      "step": 12245
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015299484026992343,
+      "loss": 0.007,
+      "step": 12246
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015297545874162262,
+      "loss": 0.0036,
+      "step": 12247
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001529560771636261,
+      "loss": 0.0183,
+      "step": 12248
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015293669553625747,
+      "loss": 0.0034,
+      "step": 12249
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015291731385984046,
+      "loss": 0.0088,
+      "step": 12250
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001528979321346988,
+      "loss": 0.0113,
+      "step": 12251
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015287855036115623,
+      "loss": 0.0021,
+      "step": 12252
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015285916853953645,
+      "loss": 0.0038,
+      "step": 12253
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015283978667016314,
+      "loss": 0.0139,
+      "step": 12254
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015282040475336,
+      "loss": 0.0031,
+      "step": 12255
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015280102278945082,
+      "loss": 0.0015,
+      "step": 12256
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015278164077875924,
+      "loss": 0.005,
+      "step": 12257
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015276225872160902,
+      "loss": 0.0099,
+      "step": 12258
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001527428766183239,
+      "loss": 0.0131,
+      "step": 12259
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001527234944692275,
+      "loss": 0.0047,
+      "step": 12260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015270411227464366,
+      "loss": 0.0019,
+      "step": 12261
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015268473003489602,
+      "loss": 0.0125,
+      "step": 12262
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015266534775030833,
+      "loss": 0.0065,
+      "step": 12263
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015264596542120427,
+      "loss": 0.0074,
+      "step": 12264
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015262658304790761,
+      "loss": 0.0069,
+      "step": 12265
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015260720063074208,
+      "loss": 0.0043,
+      "step": 12266
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015258781817003135,
+      "loss": 0.0075,
+      "step": 12267
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015256843566609915,
+      "loss": 0.0047,
+      "step": 12268
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015254905311926924,
+      "loss": 0.0361,
+      "step": 12269
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015252967052986535,
+      "loss": 0.0183,
+      "step": 12270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015251028789821123,
+      "loss": 0.0054,
+      "step": 12271
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001524909052246305,
+      "loss": 0.0033,
+      "step": 12272
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015247152250944692,
+      "loss": 0.0201,
+      "step": 12273
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015245213975298433,
+      "loss": 0.0072,
+      "step": 12274
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015243275695556632,
+      "loss": 0.0009,
+      "step": 12275
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001524133741175167,
+      "loss": 0.0025,
+      "step": 12276
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015239399123915916,
+      "loss": 0.005,
+      "step": 12277
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015237460832081748,
+      "loss": 0.011,
+      "step": 12278
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001523552253628153,
+      "loss": 0.0076,
+      "step": 12279
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015233584236547647,
+      "loss": 0.0024,
+      "step": 12280
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015231645932912467,
+      "loss": 0.0027,
+      "step": 12281
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015229707625408358,
+      "loss": 0.0048,
+      "step": 12282
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000152277693140677,
+      "loss": 0.0156,
+      "step": 12283
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015225830998922865,
+      "loss": 0.0137,
+      "step": 12284
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001522389268000623,
+      "loss": 0.0091,
+      "step": 12285
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015221954357350155,
+      "loss": 0.0143,
+      "step": 12286
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001522001603098703,
+      "loss": 0.0178,
+      "step": 12287
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001521807770094922,
+      "loss": 0.007,
+      "step": 12288
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000152161393672691,
+      "loss": 0.0056,
+      "step": 12289
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015214201029979047,
+      "loss": 0.0046,
+      "step": 12290
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015212262689111433,
+      "loss": 0.0015,
+      "step": 12291
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001521032434469863,
+      "loss": 0.0109,
+      "step": 12292
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001520838599677301,
+      "loss": 0.0105,
+      "step": 12293
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015206447645366953,
+      "loss": 0.012,
+      "step": 12294
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015204509290512835,
+      "loss": 0.0024,
+      "step": 12295
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015202570932243021,
+      "loss": 0.0043,
+      "step": 12296
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015200632570589888,
+      "loss": 0.01,
+      "step": 12297
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015198694205585816,
+      "loss": 0.0039,
+      "step": 12298
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015196755837263177,
+      "loss": 0.0018,
+      "step": 12299
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001519481746565434,
+      "loss": 0.0055,
+      "step": 12300
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001519287909079169,
+      "loss": 0.0031,
+      "step": 12301
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015190940712707588,
+      "loss": 0.0267,
+      "step": 12302
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015189002331434417,
+      "loss": 0.0223,
+      "step": 12303
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015187063947004557,
+      "loss": 0.0115,
+      "step": 12304
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015185125559450367,
+      "loss": 0.0039,
+      "step": 12305
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015183187168804236,
+      "loss": 0.0066,
+      "step": 12306
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015181248775098533,
+      "loss": 0.0061,
+      "step": 12307
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001517931037836563,
+      "loss": 0.0097,
+      "step": 12308
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015177371978637908,
+      "loss": 0.0078,
+      "step": 12309
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015175433575947736,
+      "loss": 0.0117,
+      "step": 12310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015173495170327497,
+      "loss": 0.0083,
+      "step": 12311
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015171556761809559,
+      "loss": 0.0102,
+      "step": 12312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015169618350426302,
+      "loss": 0.0103,
+      "step": 12313
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015167679936210096,
+      "loss": 0.0081,
+      "step": 12314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001516574151919332,
+      "loss": 0.0042,
+      "step": 12315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001516380309940834,
+      "loss": 0.0098,
+      "step": 12316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015161864676887547,
+      "loss": 0.0083,
+      "step": 12317
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015159926251663306,
+      "loss": 0.0069,
+      "step": 12318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015157987823767995,
+      "loss": 0.0187,
+      "step": 12319
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015156049393233987,
+      "loss": 0.0208,
+      "step": 12320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001515411096009366,
+      "loss": 0.0114,
+      "step": 12321
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015152172524379398,
+      "loss": 0.0099,
+      "step": 12322
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001515023408612356,
+      "loss": 0.0017,
+      "step": 12323
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015148295645358526,
+      "loss": 0.0067,
+      "step": 12324
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015146357202116683,
+      "loss": 0.0112,
+      "step": 12325
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015144418756430391,
+      "loss": 0.0098,
+      "step": 12326
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015142480308332038,
+      "loss": 0.0068,
+      "step": 12327
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001514054185785399,
+      "loss": 0.0181,
+      "step": 12328
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001513860340502863,
+      "loss": 0.0199,
+      "step": 12329
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001513666494988833,
+      "loss": 0.0046,
+      "step": 12330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001513472649246547,
+      "loss": 0.0092,
+      "step": 12331
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015132788032792428,
+      "loss": 0.0022,
+      "step": 12332
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015130849570901567,
+      "loss": 0.0071,
+      "step": 12333
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015128911106825272,
+      "loss": 0.0057,
+      "step": 12334
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015126972640595922,
+      "loss": 0.0061,
+      "step": 12335
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015125034172245891,
+      "loss": 0.0033,
+      "step": 12336
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001512309570180755,
+      "loss": 0.0031,
+      "step": 12337
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015121157229313278,
+      "loss": 0.0061,
+      "step": 12338
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015119218754795446,
+      "loss": 0.0073,
+      "step": 12339
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015117280278286442,
+      "loss": 0.0098,
+      "step": 12340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015115341799818636,
+      "loss": 0.0236,
+      "step": 12341
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015113403319424405,
+      "loss": 0.0058,
+      "step": 12342
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015111464837136123,
+      "loss": 0.0026,
+      "step": 12343
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015109526352986169,
+      "loss": 0.009,
+      "step": 12344
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001510758786700692,
+      "loss": 0.0034,
+      "step": 12345
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015105649379230744,
+      "loss": 0.0055,
+      "step": 12346
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001510371088969003,
+      "loss": 0.0213,
+      "step": 12347
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015101772398417148,
+      "loss": 0.0065,
+      "step": 12348
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001509983390544447,
+      "loss": 0.0077,
+      "step": 12349
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001509789541080438,
+      "loss": 0.0101,
+      "step": 12350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015095956914529252,
+      "loss": 0.0055,
+      "step": 12351
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015094018416651463,
+      "loss": 0.0105,
+      "step": 12352
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015092079917203386,
+      "loss": 0.0176,
+      "step": 12353
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015090141416217406,
+      "loss": 0.0111,
+      "step": 12354
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001508820291372589,
+      "loss": 0.0303,
+      "step": 12355
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015086264409761222,
+      "loss": 0.0103,
+      "step": 12356
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015084325904355771,
+      "loss": 0.004,
+      "step": 12357
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001508238739754192,
+      "loss": 0.014,
+      "step": 12358
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015080448889352046,
+      "loss": 0.004,
+      "step": 12359
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001507851037981852,
+      "loss": 0.0046,
+      "step": 12360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015076571868973725,
+      "loss": 0.0104,
+      "step": 12361
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015074633356850037,
+      "loss": 0.0055,
+      "step": 12362
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001507269484347983,
+      "loss": 0.0092,
+      "step": 12363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015070756328895485,
+      "loss": 0.0227,
+      "step": 12364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015068817813129368,
+      "loss": 0.0047,
+      "step": 12365
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015066879296213867,
+      "loss": 0.0393,
+      "step": 12366
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015064940778181357,
+      "loss": 0.0071,
+      "step": 12367
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001506300225906421,
+      "loss": 0.0138,
+      "step": 12368
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001506106373889481,
+      "loss": 0.0123,
+      "step": 12369
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015059125217705527,
+      "loss": 0.0634,
+      "step": 12370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015057186695528743,
+      "loss": 0.0043,
+      "step": 12371
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015055248172396836,
+      "loss": 0.0019,
+      "step": 12372
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001505330964834218,
+      "loss": 0.0169,
+      "step": 12373
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015051371123397147,
+      "loss": 0.018,
+      "step": 12374
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015049432597594124,
+      "loss": 0.0047,
+      "step": 12375
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001504749407096548,
+      "loss": 0.0122,
+      "step": 12376
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015045555543543598,
+      "loss": 0.0079,
+      "step": 12377
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001504361701536085,
+      "loss": 0.0019,
+      "step": 12378
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001504167848644962,
+      "loss": 0.0107,
+      "step": 12379
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015039739956842273,
+      "loss": 0.0076,
+      "step": 12380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000150378014265712,
+      "loss": 0.0082,
+      "step": 12381
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001503586289566877,
+      "loss": 0.0404,
+      "step": 12382
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001503392436416736,
+      "loss": 0.0034,
+      "step": 12383
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001503198583209935,
+      "loss": 0.0087,
+      "step": 12384
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001503004729949712,
+      "loss": 0.0091,
+      "step": 12385
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015028108766393042,
+      "loss": 0.002,
+      "step": 12386
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015026170232819493,
+      "loss": 0.0129,
+      "step": 12387
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015024231698808853,
+      "loss": 0.0103,
+      "step": 12388
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015022293164393497,
+      "loss": 0.0076,
+      "step": 12389
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015020354629605802,
+      "loss": 0.0423,
+      "step": 12390
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015018416094478147,
+      "loss": 0.017,
+      "step": 12391
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001501647755904291,
+      "loss": 0.0025,
+      "step": 12392
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015014539023332462,
+      "loss": 0.0138,
+      "step": 12393
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015012600487379187,
+      "loss": 0.0031,
+      "step": 12394
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015010661951215463,
+      "loss": 0.0175,
+      "step": 12395
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015008723414873667,
+      "loss": 0.0148,
+      "step": 12396
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015006784878386165,
+      "loss": 0.0064,
+      "step": 12397
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015004846341785347,
+      "loss": 0.0055,
+      "step": 12398
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015002907805103583,
+      "loss": 0.0038,
+      "step": 12399
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015000969268373258,
+      "loss": 0.0069,
+      "step": 12400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014999030731626742,
+      "loss": 0.0065,
+      "step": 12401
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014997092194896414,
+      "loss": 0.017,
+      "step": 12402
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014995153658214656,
+      "loss": 0.0155,
+      "step": 12403
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014993215121613833,
+      "loss": 0.0088,
+      "step": 12404
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014991276585126339,
+      "loss": 0.0102,
+      "step": 12405
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014989338048784534,
+      "loss": 0.0016,
+      "step": 12406
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001498739951262081,
+      "loss": 0.0039,
+      "step": 12407
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001498546097666754,
+      "loss": 0.0138,
+      "step": 12408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014983522440957091,
+      "loss": 0.0385,
+      "step": 12409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014981583905521853,
+      "loss": 0.0078,
+      "step": 12410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014979645370394198,
+      "loss": 0.0171,
+      "step": 12411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014977706835606503,
+      "loss": 0.0012,
+      "step": 12412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001497576830119115,
+      "loss": 0.0044,
+      "step": 12413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014973829767180505,
+      "loss": 0.0238,
+      "step": 12414
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014971891233606959,
+      "loss": 0.0122,
+      "step": 12415
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014969952700502877,
+      "loss": 0.0031,
+      "step": 12416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014968014167900646,
+      "loss": 0.0036,
+      "step": 12417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001496607563583264,
+      "loss": 0.0069,
+      "step": 12418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014964137104331229,
+      "loss": 0.0065,
+      "step": 12419
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014962198573428801,
+      "loss": 0.0061,
+      "step": 12420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014960260043157722,
+      "loss": 0.0083,
+      "step": 12421
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001495832151355038,
+      "loss": 0.0045,
+      "step": 12422
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001495638298463915,
+      "loss": 0.0015,
+      "step": 12423
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000149544444564564,
+      "loss": 0.0043,
+      "step": 12424
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001495250592903452,
+      "loss": 0.0101,
+      "step": 12425
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001495056740240588,
+      "loss": 0.011,
+      "step": 12426
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001494862887660285,
+      "loss": 0.0024,
+      "step": 12427
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014946690351657824,
+      "loss": 0.0101,
+      "step": 12428
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001494475182760316,
+      "loss": 0.0015,
+      "step": 12429
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014942813304471254,
+      "loss": 0.0057,
+      "step": 12430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001494087478229447,
+      "loss": 0.0299,
+      "step": 12431
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001493893626110519,
+      "loss": 0.0123,
+      "step": 12432
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014936997740935788,
+      "loss": 0.007,
+      "step": 12433
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001493505922181864,
+      "loss": 0.0099,
+      "step": 12434
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014933120703786133,
+      "loss": 0.0088,
+      "step": 12435
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014931182186870635,
+      "loss": 0.0103,
+      "step": 12436
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014929243671104518,
+      "loss": 0.003,
+      "step": 12437
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001492730515652017,
+      "loss": 0.0112,
+      "step": 12438
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001492536664314996,
+      "loss": 0.0081,
+      "step": 12439
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014923428131026272,
+      "loss": 0.0178,
+      "step": 12440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014921489620181477,
+      "loss": 0.0051,
+      "step": 12441
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014919551110647954,
+      "loss": 0.0052,
+      "step": 12442
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014917612602458077,
+      "loss": 0.0239,
+      "step": 12443
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014915674095644226,
+      "loss": 0.01,
+      "step": 12444
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014913735590238778,
+      "loss": 0.0114,
+      "step": 12445
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014911797086274113,
+      "loss": 0.0179,
+      "step": 12446
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001490985858378259,
+      "loss": 0.0038,
+      "step": 12447
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014907920082796611,
+      "loss": 0.0157,
+      "step": 12448
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014905981583348535,
+      "loss": 0.0076,
+      "step": 12449
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014904043085470746,
+      "loss": 0.0043,
+      "step": 12450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014902104589195622,
+      "loss": 0.0081,
+      "step": 12451
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001490016609455553,
+      "loss": 0.0082,
+      "step": 12452
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014898227601582855,
+      "loss": 0.0015,
+      "step": 12453
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014896289110309972,
+      "loss": 0.0064,
+      "step": 12454
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014894350620769253,
+      "loss": 0.0036,
+      "step": 12455
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014892412132993084,
+      "loss": 0.0047,
+      "step": 12456
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001489047364701383,
+      "loss": 0.015,
+      "step": 12457
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014888535162863874,
+      "loss": 0.0082,
+      "step": 12458
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014886596680575596,
+      "loss": 0.0099,
+      "step": 12459
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001488465820018136,
+      "loss": 0.0029,
+      "step": 12460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014882719721713556,
+      "loss": 0.0092,
+      "step": 12461
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001488078124520455,
+      "loss": 0.0022,
+      "step": 12462
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014878842770686722,
+      "loss": 0.0204,
+      "step": 12463
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014876904298192454,
+      "loss": 0.0024,
+      "step": 12464
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001487496582775411,
+      "loss": 0.0023,
+      "step": 12465
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014873027359404075,
+      "loss": 0.0041,
+      "step": 12466
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001487108889317472,
+      "loss": 0.0029,
+      "step": 12467
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001486915042909843,
+      "loss": 0.0106,
+      "step": 12468
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014867211967207577,
+      "loss": 0.0046,
+      "step": 12469
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014865273507534525,
+      "loss": 0.0168,
+      "step": 12470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014863335050111667,
+      "loss": 0.0082,
+      "step": 12471
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014861396594971366,
+      "loss": 0.0063,
+      "step": 12472
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001485945814214601,
+      "loss": 0.0008,
+      "step": 12473
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014857519691667965,
+      "loss": 0.0049,
+      "step": 12474
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014855581243569606,
+      "loss": 0.0051,
+      "step": 12475
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001485364279788332,
+      "loss": 0.0059,
+      "step": 12476
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014851704354641474,
+      "loss": 0.0055,
+      "step": 12477
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014849765913876441,
+      "loss": 0.0035,
+      "step": 12478
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014847827475620605,
+      "loss": 0.0109,
+      "step": 12479
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001484588903990633,
+      "loss": 0.0029,
+      "step": 12480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001484395060676601,
+      "loss": 0.0064,
+      "step": 12481
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014842012176232003,
+      "loss": 0.0029,
+      "step": 12482
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014840073748336694,
+      "loss": 0.011,
+      "step": 12483
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014838135323112453,
+      "loss": 0.0129,
+      "step": 12484
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014836196900591656,
+      "loss": 0.0099,
+      "step": 12485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001483425848080668,
+      "loss": 0.0079,
+      "step": 12486
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014832320063789907,
+      "loss": 0.0117,
+      "step": 12487
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014830381649573695,
+      "loss": 0.0113,
+      "step": 12488
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001482844323819044,
+      "loss": 0.0037,
+      "step": 12489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014826504829672498,
+      "loss": 0.0008,
+      "step": 12490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001482456642405226,
+      "loss": 0.0264,
+      "step": 12491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014822628021362094,
+      "loss": 0.0048,
+      "step": 12492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001482068962163437,
+      "loss": 0.0047,
+      "step": 12493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014818751224901467,
+      "loss": 0.0035,
+      "step": 12494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014816812831195764,
+      "loss": 0.0063,
+      "step": 12495
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001481487444054963,
+      "loss": 0.0167,
+      "step": 12496
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014812936052995446,
+      "loss": 0.0215,
+      "step": 12497
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014810997668565577,
+      "loss": 0.0082,
+      "step": 12498
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014809059287292412,
+      "loss": 0.0032,
+      "step": 12499
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014807120909208316,
+      "loss": 0.0147,
+      "step": 12500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014805182534345657,
+      "loss": 0.0076,
+      "step": 12501
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014803244162736826,
+      "loss": 0.0174,
+      "step": 12502
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014801305794414182,
+      "loss": 0.0143,
+      "step": 12503
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014799367429410112,
+      "loss": 0.0077,
+      "step": 12504
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001479742906775698,
+      "loss": 0.005,
+      "step": 12505
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014795490709487163,
+      "loss": 0.0126,
+      "step": 12506
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014793552354633045,
+      "loss": 0.0066,
+      "step": 12507
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014791614003226985,
+      "loss": 0.0118,
+      "step": 12508
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001478967565530137,
+      "loss": 0.0035,
+      "step": 12509
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001478773731088857,
+      "loss": 0.0138,
+      "step": 12510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014785798970020953,
+      "loss": 0.0164,
+      "step": 12511
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014783860632730902,
+      "loss": 0.0079,
+      "step": 12512
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014781922299050777,
+      "loss": 0.0011,
+      "step": 12513
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001477998396901297,
+      "loss": 0.0038,
+      "step": 12514
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014778045642649846,
+      "loss": 0.0181,
+      "step": 12515
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001477610731999377,
+      "loss": 0.003,
+      "step": 12516
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014774169001077135,
+      "loss": 0.0223,
+      "step": 12517
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014772230685932295,
+      "loss": 0.0092,
+      "step": 12518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001477029237459164,
+      "loss": 0.0042,
+      "step": 12519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014768354067087536,
+      "loss": 0.0074,
+      "step": 12520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014766415763452348,
+      "loss": 0.0106,
+      "step": 12521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014764477463718466,
+      "loss": 0.0144,
+      "step": 12522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001476253916791825,
+      "loss": 0.0088,
+      "step": 12523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014760600876084082,
+      "loss": 0.0015,
+      "step": 12524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001475866258824833,
+      "loss": 0.0091,
+      "step": 12525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014756724304443365,
+      "loss": 0.0072,
+      "step": 12526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014754786024701567,
+      "loss": 0.0089,
+      "step": 12527
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014752847749055308,
+      "loss": 0.0029,
+      "step": 12528
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001475090947753695,
+      "loss": 0.0041,
+      "step": 12529
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014748971210178883,
+      "loss": 0.0144,
+      "step": 12530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001474703294701346,
+      "loss": 0.0132,
+      "step": 12531
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014745094688073073,
+      "loss": 0.0034,
+      "step": 12532
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014743156433390088,
+      "loss": 0.0102,
+      "step": 12533
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014741218182996867,
+      "loss": 0.0236,
+      "step": 12534
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014739279936925792,
+      "loss": 0.0019,
+      "step": 12535
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014737341695209239,
+      "loss": 0.0152,
+      "step": 12536
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001473540345787957,
+      "loss": 0.0168,
+      "step": 12537
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001473346522496917,
+      "loss": 0.0078,
+      "step": 12538
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014731526996510395,
+      "loss": 0.0059,
+      "step": 12539
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014729588772535634,
+      "loss": 0.0173,
+      "step": 12540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014727650553077244,
+      "loss": 0.0053,
+      "step": 12541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001472571233816761,
+      "loss": 0.0095,
+      "step": 12542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014723774127839098,
+      "loss": 0.0163,
+      "step": 12543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014721835922124074,
+      "loss": 0.0062,
+      "step": 12544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014719897721054918,
+      "loss": 0.008,
+      "step": 12545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014717959524663998,
+      "loss": 0.0011,
+      "step": 12546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014716021332983684,
+      "loss": 0.0031,
+      "step": 12547
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014714083146046355,
+      "loss": 0.0031,
+      "step": 12548
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001471214496388437,
+      "loss": 0.0121,
+      "step": 12549
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001471020678653012,
+      "loss": 0.0108,
+      "step": 12550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014708268614015957,
+      "loss": 0.0053,
+      "step": 12551
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014706330446374253,
+      "loss": 0.0164,
+      "step": 12552
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014704392283637394,
+      "loss": 0.0026,
+      "step": 12553
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014702454125837733,
+      "loss": 0.0073,
+      "step": 12554
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014700515973007657,
+      "loss": 0.0033,
+      "step": 12555
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014698577825179526,
+      "loss": 0.0098,
+      "step": 12556
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014696639682385714,
+      "loss": 0.0312,
+      "step": 12557
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014694701544658596,
+      "loss": 0.0066,
+      "step": 12558
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001469276341203053,
+      "loss": 0.002,
+      "step": 12559
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014690825284533902,
+      "loss": 0.0018,
+      "step": 12560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014688887162201078,
+      "loss": 0.017,
+      "step": 12561
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001468694904506442,
+      "loss": 0.0034,
+      "step": 12562
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014685010933156308,
+      "loss": 0.013,
+      "step": 12563
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000146830728265091,
+      "loss": 0.0037,
+      "step": 12564
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014681134725155181,
+      "loss": 0.0053,
+      "step": 12565
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014679196629126912,
+      "loss": 0.005,
+      "step": 12566
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014677258538456665,
+      "loss": 0.0111,
+      "step": 12567
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014675320453176813,
+      "loss": 0.0103,
+      "step": 12568
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014673382373319712,
+      "loss": 0.0052,
+      "step": 12569
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014671444298917753,
+      "loss": 0.0052,
+      "step": 12570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014669506230003296,
+      "loss": 0.0085,
+      "step": 12571
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014667568166608697,
+      "loss": 0.0084,
+      "step": 12572
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014665630108766348,
+      "loss": 0.0009,
+      "step": 12573
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014663692056508606,
+      "loss": 0.0169,
+      "step": 12574
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014661754009867833,
+      "loss": 0.0092,
+      "step": 12575
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014659815968876417,
+      "loss": 0.0412,
+      "step": 12576
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014657877933566706,
+      "loss": 0.0076,
+      "step": 12577
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014655939903971085,
+      "loss": 0.0021,
+      "step": 12578
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014654001880121917,
+      "loss": 0.0017,
+      "step": 12579
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014652063862051568,
+      "loss": 0.0071,
+      "step": 12580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014650125849792416,
+      "loss": 0.0051,
+      "step": 12581
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014648187843376814,
+      "loss": 0.0065,
+      "step": 12582
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014646249842837147,
+      "loss": 0.0147,
+      "step": 12583
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014644311848205775,
+      "loss": 0.0065,
+      "step": 12584
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014642373859515058,
+      "loss": 0.0086,
+      "step": 12585
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014640435876797381,
+      "loss": 0.0049,
+      "step": 12586
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014638497900085098,
+      "loss": 0.0081,
+      "step": 12587
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014636559929410585,
+      "loss": 0.0026,
+      "step": 12588
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001463462196480621,
+      "loss": 0.0009,
+      "step": 12589
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001463268400630433,
+      "loss": 0.0038,
+      "step": 12590
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001463074605393733,
+      "loss": 0.0017,
+      "step": 12591
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001462880810773756,
+      "loss": 0.0191,
+      "step": 12592
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014626870167737402,
+      "loss": 0.0073,
+      "step": 12593
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014624932233969217,
+      "loss": 0.0042,
+      "step": 12594
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014622994306465363,
+      "loss": 0.0115,
+      "step": 12595
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014621056385258224,
+      "loss": 0.0035,
+      "step": 12596
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001461911847038015,
+      "loss": 0.0313,
+      "step": 12597
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014617180561863526,
+      "loss": 0.0171,
+      "step": 12598
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014615242659740706,
+      "loss": 0.0004,
+      "step": 12599
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014613304764044053,
+      "loss": 0.0162,
+      "step": 12600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014611366874805952,
+      "loss": 0.0064,
+      "step": 12601
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014609428992058755,
+      "loss": 0.0135,
+      "step": 12602
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014607491115834823,
+      "loss": 0.0164,
+      "step": 12603
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014605553246166539,
+      "loss": 0.0046,
+      "step": 12604
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014603615383086254,
+      "loss": 0.0073,
+      "step": 12605
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014601677526626346,
+      "loss": 0.0031,
+      "step": 12606
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014599739676819174,
+      "loss": 0.0084,
+      "step": 12607
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000145978018336971,
+      "loss": 0.0095,
+      "step": 12608
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014595863997292498,
+      "loss": 0.0068,
+      "step": 12609
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001459392616763773,
+      "loss": 0.0089,
+      "step": 12610
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014591988344765162,
+      "loss": 0.0014,
+      "step": 12611
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014590050528707162,
+      "loss": 0.0014,
+      "step": 12612
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014588112719496086,
+      "loss": 0.0051,
+      "step": 12613
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014586174917164308,
+      "loss": 0.0084,
+      "step": 12614
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014584237121744185,
+      "loss": 0.0062,
+      "step": 12615
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014582299333268096,
+      "loss": 0.0207,
+      "step": 12616
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014580361551768395,
+      "loss": 0.0006,
+      "step": 12617
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001457842377727744,
+      "loss": 0.01,
+      "step": 12618
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001457648600982761,
+      "loss": 0.0053,
+      "step": 12619
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014574548249451263,
+      "loss": 0.0042,
+      "step": 12620
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001457261049618076,
+      "loss": 0.0126,
+      "step": 12621
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014570672750048474,
+      "loss": 0.0045,
+      "step": 12622
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014568735011086757,
+      "loss": 0.0085,
+      "step": 12623
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014566797279327986,
+      "loss": 0.0026,
+      "step": 12624
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014564859554804516,
+      "loss": 0.002,
+      "step": 12625
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014562921837548708,
+      "loss": 0.0196,
+      "step": 12626
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001456098412759294,
+      "loss": 0.0059,
+      "step": 12627
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014559046424969554,
+      "loss": 0.0027,
+      "step": 12628
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014557108729710934,
+      "loss": 0.0069,
+      "step": 12629
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001455517104184943,
+      "loss": 0.022,
+      "step": 12630
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014553233361417408,
+      "loss": 0.012,
+      "step": 12631
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014551295688447236,
+      "loss": 0.0109,
+      "step": 12632
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014549358022971268,
+      "loss": 0.0053,
+      "step": 12633
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001454742036502188,
+      "loss": 0.0072,
+      "step": 12634
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014545482714631424,
+      "loss": 0.006,
+      "step": 12635
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014543545071832257,
+      "loss": 0.0019,
+      "step": 12636
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014541607436656758,
+      "loss": 0.0036,
+      "step": 12637
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014539669809137271,
+      "loss": 0.0029,
+      "step": 12638
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014537732189306177,
+      "loss": 0.0034,
+      "step": 12639
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001453579457719582,
+      "loss": 0.0161,
+      "step": 12640
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001453385697283857,
+      "loss": 0.0093,
+      "step": 12641
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014531919376266795,
+      "loss": 0.0138,
+      "step": 12642
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014529981787512843,
+      "loss": 0.0025,
+      "step": 12643
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001452804420660909,
+      "loss": 0.0009,
+      "step": 12644
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014526106633587886,
+      "loss": 0.0034,
+      "step": 12645
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014524169068481593,
+      "loss": 0.0022,
+      "step": 12646
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001452223151132258,
+      "loss": 0.0099,
+      "step": 12647
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000145202939621432,
+      "loss": 0.0131,
+      "step": 12648
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014518356420975816,
+      "loss": 0.0046,
+      "step": 12649
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014516418887852787,
+      "loss": 0.0069,
+      "step": 12650
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014514481362806476,
+      "loss": 0.0106,
+      "step": 12651
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014512543845869244,
+      "loss": 0.0027,
+      "step": 12652
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014510606337073454,
+      "loss": 0.0032,
+      "step": 12653
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014508668836451453,
+      "loss": 0.0028,
+      "step": 12654
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001450673134403562,
+      "loss": 0.007,
+      "step": 12655
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014504793859858293,
+      "loss": 0.0151,
+      "step": 12656
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014502856383951854,
+      "loss": 0.006,
+      "step": 12657
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001450091891634865,
+      "loss": 0.0036,
+      "step": 12658
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014498981457081037,
+      "loss": 0.0135,
+      "step": 12659
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001449704400618138,
+      "loss": 0.0079,
+      "step": 12660
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001449510656368204,
+      "loss": 0.0033,
+      "step": 12661
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014493169129615373,
+      "loss": 0.0435,
+      "step": 12662
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001449123170401374,
+      "loss": 0.0177,
+      "step": 12663
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014489294286909493,
+      "loss": 0.0134,
+      "step": 12664
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014487356878335002,
+      "loss": 0.0087,
+      "step": 12665
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014485419478322612,
+      "loss": 0.0085,
+      "step": 12666
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014483482086904696,
+      "loss": 0.0033,
+      "step": 12667
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014481544704113603,
+      "loss": 0.0017,
+      "step": 12668
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014479607329981683,
+      "loss": 0.0041,
+      "step": 12669
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014477669964541314,
+      "loss": 0.016,
+      "step": 12670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001447573260782483,
+      "loss": 0.0047,
+      "step": 12671
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001447379525986461,
+      "loss": 0.0097,
+      "step": 12672
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014471857920693004,
+      "loss": 0.0019,
+      "step": 12673
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001446992059034236,
+      "loss": 0.0228,
+      "step": 12674
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001446798326884505,
+      "loss": 0.0046,
+      "step": 12675
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014466045956233422,
+      "loss": 0.01,
+      "step": 12676
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014464108652539832,
+      "loss": 0.0201,
+      "step": 12677
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001446217135779664,
+      "loss": 0.004,
+      "step": 12678
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000144602340720362,
+      "loss": 0.0101,
+      "step": 12679
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014458296795290875,
+      "loss": 0.0023,
+      "step": 12680
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014456359527593012,
+      "loss": 0.0086,
+      "step": 12681
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014454422268974973,
+      "loss": 0.0204,
+      "step": 12682
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001445248501946911,
+      "loss": 0.012,
+      "step": 12683
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014450547779107777,
+      "loss": 0.0037,
+      "step": 12684
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001444861054792334,
+      "loss": 0.0046,
+      "step": 12685
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014446673325948148,
+      "loss": 0.0061,
+      "step": 12686
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001444473611321455,
+      "loss": 0.0062,
+      "step": 12687
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014442798909754915,
+      "loss": 0.0045,
+      "step": 12688
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001444086171560158,
+      "loss": 0.0125,
+      "step": 12689
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001443892453078692,
+      "loss": 0.0164,
+      "step": 12690
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014436987355343273,
+      "loss": 0.0043,
+      "step": 12691
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014435050189303002,
+      "loss": 0.0035,
+      "step": 12692
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001443311303269846,
+      "loss": 0.0145,
+      "step": 12693
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014431175885561994,
+      "loss": 0.014,
+      "step": 12694
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014429238747925972,
+      "loss": 0.0127,
+      "step": 12695
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001442730161982274,
+      "loss": 0.0032,
+      "step": 12696
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014425364501284647,
+      "loss": 0.0108,
+      "step": 12697
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014423427392344057,
+      "loss": 0.0111,
+      "step": 12698
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001442149029303332,
+      "loss": 0.0078,
+      "step": 12699
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001441955320338478,
+      "loss": 0.008,
+      "step": 12700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014417616123430798,
+      "loss": 0.0093,
+      "step": 12701
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014415679053203726,
+      "loss": 0.0042,
+      "step": 12702
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001441374199273592,
+      "loss": 0.0059,
+      "step": 12703
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001441180494205973,
+      "loss": 0.0048,
+      "step": 12704
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014409867901207503,
+      "loss": 0.016,
+      "step": 12705
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000144079308702116,
+      "loss": 0.0034,
+      "step": 12706
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014405993849104365,
+      "loss": 0.0076,
+      "step": 12707
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001440405683791816,
+      "loss": 0.0024,
+      "step": 12708
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001440211983668533,
+      "loss": 0.0057,
+      "step": 12709
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001440018284543822,
+      "loss": 0.0119,
+      "step": 12710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000143982458642092,
+      "loss": 0.0113,
+      "step": 12711
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014396308893030598,
+      "loss": 0.0102,
+      "step": 12712
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014394371931934782,
+      "loss": 0.0052,
+      "step": 12713
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000143924349809541,
+      "loss": 0.0044,
+      "step": 12714
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014390498040120896,
+      "loss": 0.009,
+      "step": 12715
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014388561109467533,
+      "loss": 0.0013,
+      "step": 12716
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014386624189026344,
+      "loss": 0.0027,
+      "step": 12717
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014384687278829698,
+      "loss": 0.0083,
+      "step": 12718
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014382750378909932,
+      "loss": 0.0135,
+      "step": 12719
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014380813489299398,
+      "loss": 0.0224,
+      "step": 12720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014378876610030453,
+      "loss": 0.0067,
+      "step": 12721
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014376939741135437,
+      "loss": 0.021,
+      "step": 12722
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014375002882646704,
+      "loss": 0.0072,
+      "step": 12723
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000143730660345966,
+      "loss": 0.0089,
+      "step": 12724
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014371129197017477,
+      "loss": 0.0028,
+      "step": 12725
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014369192369941686,
+      "loss": 0.005,
+      "step": 12726
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014367255553401578,
+      "loss": 0.0034,
+      "step": 12727
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014365318747429487,
+      "loss": 0.0018,
+      "step": 12728
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014363381952057778,
+      "loss": 0.0031,
+      "step": 12729
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014361445167318786,
+      "loss": 0.0072,
+      "step": 12730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001435950839324487,
+      "loss": 0.0053,
+      "step": 12731
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001435757162986837,
+      "loss": 0.0104,
+      "step": 12732
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014355634877221638,
+      "loss": 0.0066,
+      "step": 12733
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001435369813533702,
+      "loss": 0.0106,
+      "step": 12734
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014351761404246863,
+      "loss": 0.0275,
+      "step": 12735
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014349824683983513,
+      "loss": 0.0036,
+      "step": 12736
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014347887974579322,
+      "loss": 0.0169,
+      "step": 12737
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014345951276066625,
+      "loss": 0.0029,
+      "step": 12738
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014344014588477783,
+      "loss": 0.0134,
+      "step": 12739
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001434207791184513,
+      "loss": 0.0128,
+      "step": 12740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014340141246201024,
+      "loss": 0.0351,
+      "step": 12741
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014338204591577805,
+      "loss": 0.0034,
+      "step": 12742
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014336267948007815,
+      "loss": 0.0049,
+      "step": 12743
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014334331315523402,
+      "loss": 0.0028,
+      "step": 12744
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014332394694156913,
+      "loss": 0.0127,
+      "step": 12745
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014330458083940695,
+      "loss": 0.0072,
+      "step": 12746
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014328521484907093,
+      "loss": 0.0126,
+      "step": 12747
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014326584897088443,
+      "loss": 0.0259,
+      "step": 12748
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014324648320517103,
+      "loss": 0.0086,
+      "step": 12749
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014322711755225412,
+      "loss": 0.0228,
+      "step": 12750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014320775201245704,
+      "loss": 0.0052,
+      "step": 12751
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001431883865861034,
+      "loss": 0.011,
+      "step": 12752
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001431690212735165,
+      "loss": 0.0078,
+      "step": 12753
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014314965607501988,
+      "loss": 0.0123,
+      "step": 12754
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014313029099093696,
+      "loss": 0.0035,
+      "step": 12755
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014311092602159106,
+      "loss": 0.004,
+      "step": 12756
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014309156116730582,
+      "loss": 0.0116,
+      "step": 12757
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014307219642840444,
+      "loss": 0.0108,
+      "step": 12758
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014305283180521052,
+      "loss": 0.0201,
+      "step": 12759
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014303346729804744,
+      "loss": 0.0352,
+      "step": 12760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014301410290723854,
+      "loss": 0.0092,
+      "step": 12761
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014299473863310736,
+      "loss": 0.0146,
+      "step": 12762
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001429753744759772,
+      "loss": 0.0105,
+      "step": 12763
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014295601043617164,
+      "loss": 0.0148,
+      "step": 12764
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014293664651401396,
+      "loss": 0.0063,
+      "step": 12765
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014291728270982758,
+      "loss": 0.0016,
+      "step": 12766
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000142897919023936,
+      "loss": 0.002,
+      "step": 12767
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014287855545666253,
+      "loss": 0.0059,
+      "step": 12768
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014285919200833068,
+      "loss": 0.0117,
+      "step": 12769
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014283982867926382,
+      "loss": 0.0118,
+      "step": 12770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014282046546978527,
+      "loss": 0.0012,
+      "step": 12771
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014280110238021855,
+      "loss": 0.0176,
+      "step": 12772
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014278173941088698,
+      "loss": 0.0245,
+      "step": 12773
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014276237656211398,
+      "loss": 0.0093,
+      "step": 12774
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014274301383422296,
+      "loss": 0.0093,
+      "step": 12775
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014272365122753732,
+      "loss": 0.003,
+      "step": 12776
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014270428874238044,
+      "loss": 0.019,
+      "step": 12777
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001426849263790757,
+      "loss": 0.0279,
+      "step": 12778
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014266556413794647,
+      "loss": 0.0133,
+      "step": 12779
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014264620201931622,
+      "loss": 0.0064,
+      "step": 12780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014262684002350822,
+      "loss": 0.0161,
+      "step": 12781
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014260747815084596,
+      "loss": 0.0038,
+      "step": 12782
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014258811640165277,
+      "loss": 0.0089,
+      "step": 12783
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014256875477625198,
+      "loss": 0.0205,
+      "step": 12784
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014254939327496704,
+      "loss": 0.0075,
+      "step": 12785
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014253003189812128,
+      "loss": 0.0069,
+      "step": 12786
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014251067064603812,
+      "loss": 0.0008,
+      "step": 12787
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001424913095190409,
+      "loss": 0.0073,
+      "step": 12788
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014247194851745292,
+      "loss": 0.0118,
+      "step": 12789
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001424525876415977,
+      "loss": 0.0051,
+      "step": 12790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014243322689179842,
+      "loss": 0.0115,
+      "step": 12791
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001424138662683786,
+      "loss": 0.0091,
+      "step": 12792
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014239450577166156,
+      "loss": 0.0085,
+      "step": 12793
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014237514540197055,
+      "loss": 0.0098,
+      "step": 12794
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001423557851596291,
+      "loss": 0.01,
+      "step": 12795
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014233642504496034,
+      "loss": 0.0091,
+      "step": 12796
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014231706505828782,
+      "loss": 0.012,
+      "step": 12797
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014229770519993482,
+      "loss": 0.0095,
+      "step": 12798
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014227834547022466,
+      "loss": 0.0096,
+      "step": 12799
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014225898586948076,
+      "loss": 0.0045,
+      "step": 12800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014223962639802636,
+      "loss": 0.0046,
+      "step": 12801
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014222026705618484,
+      "loss": 0.005,
+      "step": 12802
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014220090784427958,
+      "loss": 0.013,
+      "step": 12803
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014218154876263382,
+      "loss": 0.0116,
+      "step": 12804
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014216218981157104,
+      "loss": 0.0025,
+      "step": 12805
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014214283099141443,
+      "loss": 0.0128,
+      "step": 12806
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014212347230248737,
+      "loss": 0.0167,
+      "step": 12807
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001421041137451132,
+      "loss": 0.0033,
+      "step": 12808
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014208475531961523,
+      "loss": 0.0044,
+      "step": 12809
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014206539702631683,
+      "loss": 0.0028,
+      "step": 12810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014204603886554127,
+      "loss": 0.0037,
+      "step": 12811
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001420266808376118,
+      "loss": 0.0051,
+      "step": 12812
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014200732294285191,
+      "loss": 0.0118,
+      "step": 12813
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014198796518158471,
+      "loss": 0.0044,
+      "step": 12814
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001419686075541337,
+      "loss": 0.0159,
+      "step": 12815
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014194925006082208,
+      "loss": 0.007,
+      "step": 12816
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001419298927019732,
+      "loss": 0.0039,
+      "step": 12817
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001419105354779103,
+      "loss": 0.0106,
+      "step": 12818
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014189117838895668,
+      "loss": 0.022,
+      "step": 12819
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014187182143543582,
+      "loss": 0.0254,
+      "step": 12820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014185246461767084,
+      "loss": 0.0076,
+      "step": 12821
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014183310793598503,
+      "loss": 0.0005,
+      "step": 12822
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001418137513907018,
+      "loss": 0.0064,
+      "step": 12823
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014179439498214437,
+      "loss": 0.0177,
+      "step": 12824
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014177503871063595,
+      "loss": 0.0074,
+      "step": 12825
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014175568257649996,
+      "loss": 0.0102,
+      "step": 12826
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014173632658005963,
+      "loss": 0.0052,
+      "step": 12827
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014171697072163825,
+      "loss": 0.0104,
+      "step": 12828
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001416976150015591,
+      "loss": 0.0033,
+      "step": 12829
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014167825942014542,
+      "loss": 0.0142,
+      "step": 12830
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014165890397772058,
+      "loss": 0.0029,
+      "step": 12831
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014163954867460772,
+      "loss": 0.0045,
+      "step": 12832
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014162019351113023,
+      "loss": 0.0044,
+      "step": 12833
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014160083848761135,
+      "loss": 0.0188,
+      "step": 12834
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001415814836043742,
+      "loss": 0.0213,
+      "step": 12835
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001415621288617423,
+      "loss": 0.0066,
+      "step": 12836
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014154277426003866,
+      "loss": 0.0267,
+      "step": 12837
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014152341979958673,
+      "loss": 0.009,
+      "step": 12838
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001415040654807097,
+      "loss": 0.0099,
+      "step": 12839
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014148471130373075,
+      "loss": 0.0022,
+      "step": 12840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014146535726897326,
+      "loss": 0.0082,
+      "step": 12841
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014144600337676034,
+      "loss": 0.0071,
+      "step": 12842
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001414266496274154,
+      "loss": 0.0046,
+      "step": 12843
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014140729602126156,
+      "loss": 0.0017,
+      "step": 12844
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014138794255862206,
+      "loss": 0.0093,
+      "step": 12845
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014136858923982022,
+      "loss": 0.0136,
+      "step": 12846
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001413492360651792,
+      "loss": 0.0053,
+      "step": 12847
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001413298830350223,
+      "loss": 0.0105,
+      "step": 12848
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014131053014967268,
+      "loss": 0.0089,
+      "step": 12849
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014129117740945357,
+      "loss": 0.0005,
+      "step": 12850
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014127182481468833,
+      "loss": 0.0059,
+      "step": 12851
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001412524723657001,
+      "loss": 0.0086,
+      "step": 12852
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000141233120062812,
+      "loss": 0.0071,
+      "step": 12853
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014121376790634743,
+      "loss": 0.0059,
+      "step": 12854
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014119441589662945,
+      "loss": 0.0099,
+      "step": 12855
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001411750640339814,
+      "loss": 0.009,
+      "step": 12856
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014115571231872642,
+      "loss": 0.0019,
+      "step": 12857
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014113636075118775,
+      "loss": 0.0187,
+      "step": 12858
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014111700933168856,
+      "loss": 0.0118,
+      "step": 12859
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001410976580605521,
+      "loss": 0.0108,
+      "step": 12860
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014107830693810156,
+      "loss": 0.0251,
+      "step": 12861
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014105895596466014,
+      "loss": 0.0079,
+      "step": 12862
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014103960514055098,
+      "loss": 0.007,
+      "step": 12863
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001410202544660974,
+      "loss": 0.0081,
+      "step": 12864
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014100090394162247,
+      "loss": 0.0075,
+      "step": 12865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014098155356744946,
+      "loss": 0.0014,
+      "step": 12866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014096220334390157,
+      "loss": 0.011,
+      "step": 12867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014094285327130183,
+      "loss": 0.0183,
+      "step": 12868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014092350334997362,
+      "loss": 0.0186,
+      "step": 12869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014090415358024002,
+      "loss": 0.0081,
+      "step": 12870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001408848039624242,
+      "loss": 0.0009,
+      "step": 12871
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014086545449684942,
+      "loss": 0.0017,
+      "step": 12872
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014084610518383872,
+      "loss": 0.0059,
+      "step": 12873
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001408267560237154,
+      "loss": 0.0127,
+      "step": 12874
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014080740701680257,
+      "loss": 0.0115,
+      "step": 12875
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014078805816342336,
+      "loss": 0.0107,
+      "step": 12876
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000140768709463901,
+      "loss": 0.0166,
+      "step": 12877
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014074936091855857,
+      "loss": 0.0045,
+      "step": 12878
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014073001252771928,
+      "loss": 0.0042,
+      "step": 12879
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014071066429170633,
+      "loss": 0.0071,
+      "step": 12880
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014069131621084277,
+      "loss": 0.0223,
+      "step": 12881
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014067196828545185,
+      "loss": 0.0159,
+      "step": 12882
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001406526205158566,
+      "loss": 0.0071,
+      "step": 12883
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001406332729023803,
+      "loss": 0.0059,
+      "step": 12884
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014061392544534602,
+      "loss": 0.0032,
+      "step": 12885
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014059457814507686,
+      "loss": 0.0129,
+      "step": 12886
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014057523100189604,
+      "loss": 0.0014,
+      "step": 12887
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001405558840161266,
+      "loss": 0.006,
+      "step": 12888
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014053653718809178,
+      "loss": 0.0047,
+      "step": 12889
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014051719051811463,
+      "loss": 0.0051,
+      "step": 12890
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014049784400651825,
+      "loss": 0.0142,
+      "step": 12891
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001404784976536259,
+      "loss": 0.0093,
+      "step": 12892
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001404591514597605,
+      "loss": 0.0043,
+      "step": 12893
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001404398054252454,
+      "loss": 0.0071,
+      "step": 12894
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014042045955040358,
+      "loss": 0.0085,
+      "step": 12895
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014040111383555807,
+      "loss": 0.0112,
+      "step": 12896
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001403817682810322,
+      "loss": 0.0121,
+      "step": 12897
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014036242288714888,
+      "loss": 0.0207,
+      "step": 12898
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001403430776542313,
+      "loss": 0.0082,
+      "step": 12899
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014032373258260255,
+      "loss": 0.0033,
+      "step": 12900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014030438767258574,
+      "loss": 0.0045,
+      "step": 12901
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014028504292450396,
+      "loss": 0.0072,
+      "step": 12902
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014026569833868033,
+      "loss": 0.005,
+      "step": 12903
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014024635391543785,
+      "loss": 0.0071,
+      "step": 12904
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001402270096550997,
+      "loss": 0.0017,
+      "step": 12905
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014020766555798892,
+      "loss": 0.0175,
+      "step": 12906
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014018832162442866,
+      "loss": 0.0126,
+      "step": 12907
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014016897785474193,
+      "loss": 0.0149,
+      "step": 12908
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014014963424925178,
+      "loss": 0.0115,
+      "step": 12909
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014013029080828137,
+      "loss": 0.0096,
+      "step": 12910
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014011094753215374,
+      "loss": 0.0055,
+      "step": 12911
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014009160442119193,
+      "loss": 0.0005,
+      "step": 12912
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014007226147571907,
+      "loss": 0.0072,
+      "step": 12913
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014005291869605813,
+      "loss": 0.0059,
+      "step": 12914
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014003357608253228,
+      "loss": 0.0088,
+      "step": 12915
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014001423363546445,
+      "loss": 0.0026,
+      "step": 12916
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013999489135517785,
+      "loss": 0.0185,
+      "step": 12917
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013997554924199545,
+      "loss": 0.0178,
+      "step": 12918
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001399562072962402,
+      "loss": 0.0068,
+      "step": 12919
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013993686551823535,
+      "loss": 0.0171,
+      "step": 12920
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001399175239083038,
+      "loss": 0.0055,
+      "step": 12921
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001398981824667686,
+      "loss": 0.0064,
+      "step": 12922
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001398788411939529,
+      "loss": 0.0216,
+      "step": 12923
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013985950009017954,
+      "loss": 0.0086,
+      "step": 12924
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013984015915577178,
+      "loss": 0.0911,
+      "step": 12925
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013982081839105254,
+      "loss": 0.0027,
+      "step": 12926
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013980147779634478,
+      "loss": 0.0064,
+      "step": 12927
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013978213737197165,
+      "loss": 0.0064,
+      "step": 12928
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013976279711825604,
+      "loss": 0.0081,
+      "step": 12929
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013974345703552112,
+      "loss": 0.0068,
+      "step": 12930
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013972411712408978,
+      "loss": 0.0059,
+      "step": 12931
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013970477738428505,
+      "loss": 0.0089,
+      "step": 12932
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013968543781643004,
+      "loss": 0.0359,
+      "step": 12933
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013966609842084763,
+      "loss": 0.0136,
+      "step": 12934
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013964675919786096,
+      "loss": 0.0103,
+      "step": 12935
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013962742014779296,
+      "loss": 0.0111,
+      "step": 12936
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013960808127096651,
+      "loss": 0.0091,
+      "step": 12937
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013958874256770484,
+      "loss": 0.002,
+      "step": 12938
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001395694040383307,
+      "loss": 0.0052,
+      "step": 12939
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013955006568316732,
+      "loss": 0.0033,
+      "step": 12940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013953072750253753,
+      "loss": 0.0067,
+      "step": 12941
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013951138949676434,
+      "loss": 0.0137,
+      "step": 12942
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013949205166617073,
+      "loss": 0.0151,
+      "step": 12943
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013947271401107967,
+      "loss": 0.0178,
+      "step": 12944
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001394533765318142,
+      "loss": 0.0067,
+      "step": 12945
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001394340392286973,
+      "loss": 0.0086,
+      "step": 12946
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013941470210205177,
+      "loss": 0.0259,
+      "step": 12947
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001393953651522008,
+      "loss": 0.0065,
+      "step": 12948
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013937602837946724,
+      "loss": 0.0115,
+      "step": 12949
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013935669178417398,
+      "loss": 0.0103,
+      "step": 12950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001393373553666441,
+      "loss": 0.0107,
+      "step": 12951
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013931801912720054,
+      "loss": 0.0055,
+      "step": 12952
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001392986830661662,
+      "loss": 0.003,
+      "step": 12953
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001392793471838641,
+      "loss": 0.0017,
+      "step": 12954
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013926001148061704,
+      "loss": 0.0211,
+      "step": 12955
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013924067595674815,
+      "loss": 0.0073,
+      "step": 12956
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013922134061258022,
+      "loss": 0.0094,
+      "step": 12957
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001392020054484363,
+      "loss": 0.0086,
+      "step": 12958
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013918267046463928,
+      "loss": 0.0102,
+      "step": 12959
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013916333566151202,
+      "loss": 0.0023,
+      "step": 12960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013914400103937755,
+      "loss": 0.0076,
+      "step": 12961
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013912466659855873,
+      "loss": 0.0084,
+      "step": 12962
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013910533233937853,
+      "loss": 0.0019,
+      "step": 12963
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013908599826215987,
+      "loss": 0.0029,
+      "step": 12964
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013906666436722554,
+      "loss": 0.0017,
+      "step": 12965
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013904733065489867,
+      "loss": 0.0116,
+      "step": 12966
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013902799712550197,
+      "loss": 0.0081,
+      "step": 12967
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013900866377935847,
+      "loss": 0.0048,
+      "step": 12968
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013898933061679109,
+      "loss": 0.0182,
+      "step": 12969
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013896999763812255,
+      "loss": 0.013,
+      "step": 12970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013895066484367595,
+      "loss": 0.0099,
+      "step": 12971
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001389313322337741,
+      "loss": 0.0039,
+      "step": 12972
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013891199980873987,
+      "loss": 0.0049,
+      "step": 12973
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013889266756889618,
+      "loss": 0.0125,
+      "step": 12974
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013887333551456586,
+      "loss": 0.0045,
+      "step": 12975
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001388540036460719,
+      "loss": 0.0065,
+      "step": 12976
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013883467196373715,
+      "loss": 0.0031,
+      "step": 12977
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013881534046788439,
+      "loss": 0.0115,
+      "step": 12978
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001387960091588366,
+      "loss": 0.0187,
+      "step": 12979
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013877667803691656,
+      "loss": 0.0016,
+      "step": 12980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013875734710244722,
+      "loss": 0.0065,
+      "step": 12981
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001387380163557514,
+      "loss": 0.0068,
+      "step": 12982
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013871868579715195,
+      "loss": 0.0035,
+      "step": 12983
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013869935542697176,
+      "loss": 0.012,
+      "step": 12984
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001386800252455336,
+      "loss": 0.012,
+      "step": 12985
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001386606952531605,
+      "loss": 0.0032,
+      "step": 12986
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013864136545017514,
+      "loss": 0.0264,
+      "step": 12987
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001386220358369004,
+      "loss": 0.0078,
+      "step": 12988
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013860270641365916,
+      "loss": 0.0119,
+      "step": 12989
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001385833771807742,
+      "loss": 0.0226,
+      "step": 12990
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013856404813856845,
+      "loss": 0.0038,
+      "step": 12991
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013854471928736464,
+      "loss": 0.0026,
+      "step": 12992
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013852539062748567,
+      "loss": 0.0176,
+      "step": 12993
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001385060621592543,
+      "loss": 0.0057,
+      "step": 12994
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013848673388299347,
+      "loss": 0.0077,
+      "step": 12995
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001384674057990258,
+      "loss": 0.0044,
+      "step": 12996
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013844807790767432,
+      "loss": 0.0084,
+      "step": 12997
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013842875020926167,
+      "loss": 0.0084,
+      "step": 12998
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001384094227041108,
+      "loss": 0.0182,
+      "step": 12999
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013839009539254443,
+      "loss": 0.0026,
+      "step": 13000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013837076827488535,
+      "loss": 0.0165,
+      "step": 13001
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013835144135145643,
+      "loss": 0.0149,
+      "step": 13002
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013833211462258038,
+      "loss": 0.0019,
+      "step": 13003
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013831278808858004,
+      "loss": 0.0095,
+      "step": 13004
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013829346174977828,
+      "loss": 0.0057,
+      "step": 13005
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001382741356064977,
+      "loss": 0.0042,
+      "step": 13006
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001382548096590613,
+      "loss": 0.012,
+      "step": 13007
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013823548390779167,
+      "loss": 0.0063,
+      "step": 13008
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001382161583530117,
+      "loss": 0.0091,
+      "step": 13009
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013819683299504416,
+      "loss": 0.0053,
+      "step": 13010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001381775078342117,
+      "loss": 0.0035,
+      "step": 13011
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013815818287083725,
+      "loss": 0.0103,
+      "step": 13012
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013813885810524344,
+      "loss": 0.0061,
+      "step": 13013
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013811953353775315,
+      "loss": 0.0078,
+      "step": 13014
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013810020916868907,
+      "loss": 0.0075,
+      "step": 13015
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013808088499837388,
+      "loss": 0.022,
+      "step": 13016
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001380615610271305,
+      "loss": 0.0085,
+      "step": 13017
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013804223725528153,
+      "loss": 0.0224,
+      "step": 13018
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001380229136831498,
+      "loss": 0.0129,
+      "step": 13019
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013800359031105805,
+      "loss": 0.0053,
+      "step": 13020
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013798426713932893,
+      "loss": 0.0168,
+      "step": 13021
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013796494416828527,
+      "loss": 0.0055,
+      "step": 13022
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013794562139824975,
+      "loss": 0.0205,
+      "step": 13023
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001379262988295451,
+      "loss": 0.0086,
+      "step": 13024
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000137906976462494,
+      "loss": 0.0094,
+      "step": 13025
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013788765429741928,
+      "loss": 0.0202,
+      "step": 13026
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013786833233464358,
+      "loss": 0.003,
+      "step": 13027
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013784901057448966,
+      "loss": 0.0073,
+      "step": 13028
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013782968901728015,
+      "loss": 0.0091,
+      "step": 13029
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013781036766333785,
+      "loss": 0.0055,
+      "step": 13030
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013779104651298537,
+      "loss": 0.0078,
+      "step": 13031
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013777172556654553,
+      "loss": 0.0068,
+      "step": 13032
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013775240482434092,
+      "loss": 0.0056,
+      "step": 13033
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013773308428669426,
+      "loss": 0.0015,
+      "step": 13034
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013771376395392822,
+      "loss": 0.0051,
+      "step": 13035
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013769444382636555,
+      "loss": 0.017,
+      "step": 13036
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013767512390432888,
+      "loss": 0.0151,
+      "step": 13037
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013765580418814097,
+      "loss": 0.0028,
+      "step": 13038
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013763648467812433,
+      "loss": 0.0147,
+      "step": 13039
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013761716537460182,
+      "loss": 0.0138,
+      "step": 13040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013759784627789595,
+      "loss": 0.0266,
+      "step": 13041
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013757852738832953,
+      "loss": 0.0118,
+      "step": 13042
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013755920870622514,
+      "loss": 0.0078,
+      "step": 13043
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001375398902319054,
+      "loss": 0.0114,
+      "step": 13044
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013752057196569304,
+      "loss": 0.0062,
+      "step": 13045
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001375012539079107,
+      "loss": 0.0061,
+      "step": 13046
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013748193605888094,
+      "loss": 0.0159,
+      "step": 13047
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013746261841892656,
+      "loss": 0.0058,
+      "step": 13048
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013744330098837006,
+      "loss": 0.0119,
+      "step": 13049
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013742398376753417,
+      "loss": 0.0096,
+      "step": 13050
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013740466675674152,
+      "loss": 0.0036,
+      "step": 13051
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013738534995631465,
+      "loss": 0.0072,
+      "step": 13052
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013736603336657628,
+      "loss": 0.0091,
+      "step": 13053
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013734671698784897,
+      "loss": 0.0106,
+      "step": 13054
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013732740082045543,
+      "loss": 0.0029,
+      "step": 13055
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013730808486471817,
+      "loss": 0.0151,
+      "step": 13056
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013728876912095982,
+      "loss": 0.0187,
+      "step": 13057
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001372694535895031,
+      "loss": 0.0105,
+      "step": 13058
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013725013827067047,
+      "loss": 0.0078,
+      "step": 13059
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013723082316478464,
+      "loss": 0.0048,
+      "step": 13060
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013721150827216818,
+      "loss": 0.0097,
+      "step": 13061
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013719219359314362,
+      "loss": 0.0095,
+      "step": 13062
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013717287912803366,
+      "loss": 0.0145,
+      "step": 13063
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013715356487716075,
+      "loss": 0.0084,
+      "step": 13064
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013713425084084765,
+      "loss": 0.0071,
+      "step": 13065
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013711493701941679,
+      "loss": 0.0118,
+      "step": 13066
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001370956234131908,
+      "loss": 0.0117,
+      "step": 13067
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013707631002249227,
+      "loss": 0.0022,
+      "step": 13068
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001370569968476438,
+      "loss": 0.0065,
+      "step": 13069
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013703768388896785,
+      "loss": 0.0276,
+      "step": 13070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013701837114678709,
+      "loss": 0.0055,
+      "step": 13071
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013699905862142397,
+      "loss": 0.0064,
+      "step": 13072
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001369797463132012,
+      "loss": 0.0073,
+      "step": 13073
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013696043422244125,
+      "loss": 0.0072,
+      "step": 13074
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013694112234946657,
+      "loss": 0.0146,
+      "step": 13075
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013692181069459983,
+      "loss": 0.0114,
+      "step": 13076
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013690249925816356,
+      "loss": 0.0024,
+      "step": 13077
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013688318804048027,
+      "loss": 0.0028,
+      "step": 13078
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013686387704187254,
+      "loss": 0.0089,
+      "step": 13079
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013684456626266277,
+      "loss": 0.0068,
+      "step": 13080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013682525570317365,
+      "loss": 0.0032,
+      "step": 13081
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013680594536372756,
+      "loss": 0.0042,
+      "step": 13082
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013678663524464718,
+      "loss": 0.0041,
+      "step": 13083
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001367673253462549,
+      "loss": 0.002,
+      "step": 13084
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013674801566887322,
+      "loss": 0.0029,
+      "step": 13085
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013672870621282473,
+      "loss": 0.0134,
+      "step": 13086
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013670939697843187,
+      "loss": 0.0118,
+      "step": 13087
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001366900879660172,
+      "loss": 0.0183,
+      "step": 13088
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001366707791759032,
+      "loss": 0.019,
+      "step": 13089
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013665147060841227,
+      "loss": 0.007,
+      "step": 13090
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013663216226386706,
+      "loss": 0.0074,
+      "step": 13091
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001366128541425899,
+      "loss": 0.0025,
+      "step": 13092
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013659354624490343,
+      "loss": 0.0111,
+      "step": 13093
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013657423857113004,
+      "loss": 0.015,
+      "step": 13094
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013655493112159212,
+      "loss": 0.0186,
+      "step": 13095
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013653562389661232,
+      "loss": 0.0045,
+      "step": 13096
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013651631689651298,
+      "loss": 0.0022,
+      "step": 13097
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001364970101216166,
+      "loss": 0.0033,
+      "step": 13098
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013647770357224563,
+      "loss": 0.009,
+      "step": 13099
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001364583972487225,
+      "loss": 0.006,
+      "step": 13100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001364390911513698,
+      "loss": 0.0008,
+      "step": 13101
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013641978528050982,
+      "loss": 0.0036,
+      "step": 13102
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013640047963646502,
+      "loss": 0.0193,
+      "step": 13103
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013638117421955795,
+      "loss": 0.013,
+      "step": 13104
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001363618690301109,
+      "loss": 0.0028,
+      "step": 13105
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013634256406844646,
+      "loss": 0.0148,
+      "step": 13106
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013632325933488694,
+      "loss": 0.0146,
+      "step": 13107
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001363039548297548,
+      "loss": 0.0077,
+      "step": 13108
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001362846505533725,
+      "loss": 0.0096,
+      "step": 13109
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013626534650606237,
+      "loss": 0.0253,
+      "step": 13110
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013624604268814693,
+      "loss": 0.0042,
+      "step": 13111
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013622673909994857,
+      "loss": 0.0059,
+      "step": 13112
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013620743574178958,
+      "loss": 0.0093,
+      "step": 13113
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013618813261399252,
+      "loss": 0.0159,
+      "step": 13114
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013616882971687966,
+      "loss": 0.0037,
+      "step": 13115
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001361495270507735,
+      "loss": 0.0091,
+      "step": 13116
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013613022461599636,
+      "loss": 0.0023,
+      "step": 13117
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013611092241287066,
+      "loss": 0.0087,
+      "step": 13118
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013609162044171875,
+      "loss": 0.0063,
+      "step": 13119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013607231870286306,
+      "loss": 0.0053,
+      "step": 13120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001360530171966259,
+      "loss": 0.0202,
+      "step": 13121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013603371592332974,
+      "loss": 0.0047,
+      "step": 13122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001360144148832968,
+      "loss": 0.0039,
+      "step": 13123
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013599511407684964,
+      "loss": 0.0209,
+      "step": 13124
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013597581350431048,
+      "loss": 0.0182,
+      "step": 13125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013595651316600165,
+      "loss": 0.0088,
+      "step": 13126
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001359372130622456,
+      "loss": 0.0122,
+      "step": 13127
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001359179131933646,
+      "loss": 0.0047,
+      "step": 13128
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013589861355968105,
+      "loss": 0.0085,
+      "step": 13129
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001358793141615173,
+      "loss": 0.0208,
+      "step": 13130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013586001499919561,
+      "loss": 0.0116,
+      "step": 13131
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013584071607303843,
+      "loss": 0.0072,
+      "step": 13132
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013582141738336794,
+      "loss": 0.0097,
+      "step": 13133
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001358021189305066,
+      "loss": 0.01,
+      "step": 13134
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001357828207147767,
+      "loss": 0.0308,
+      "step": 13135
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013576352273650044,
+      "loss": 0.0067,
+      "step": 13136
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013574422499600033,
+      "loss": 0.0075,
+      "step": 13137
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013572492749359846,
+      "loss": 0.0064,
+      "step": 13138
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013570563022961732,
+      "loss": 0.0054,
+      "step": 13139
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013568633320437912,
+      "loss": 0.0031,
+      "step": 13140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013566703641820614,
+      "loss": 0.0108,
+      "step": 13141
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013564773987142077,
+      "loss": 0.0011,
+      "step": 13142
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013562844356434522,
+      "loss": 0.0088,
+      "step": 13143
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013560914749730175,
+      "loss": 0.0035,
+      "step": 13144
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013558985167061273,
+      "loss": 0.0168,
+      "step": 13145
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013557055608460036,
+      "loss": 0.0045,
+      "step": 13146
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013555126073958696,
+      "loss": 0.0029,
+      "step": 13147
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013553196563589475,
+      "loss": 0.001,
+      "step": 13148
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013551267077384604,
+      "loss": 0.0067,
+      "step": 13149
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013549337615376307,
+      "loss": 0.0059,
+      "step": 13150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013547408177596812,
+      "loss": 0.0107,
+      "step": 13151
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001354547876407834,
+      "loss": 0.0033,
+      "step": 13152
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013543549374853123,
+      "loss": 0.017,
+      "step": 13153
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013541620009953372,
+      "loss": 0.0034,
+      "step": 13154
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013539690669411327,
+      "loss": 0.0144,
+      "step": 13155
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000135377613532592,
+      "loss": 0.0075,
+      "step": 13156
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013535832061529224,
+      "loss": 0.0099,
+      "step": 13157
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013533902794253614,
+      "loss": 0.0161,
+      "step": 13158
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013531973551464592,
+      "loss": 0.0074,
+      "step": 13159
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013530044333194384,
+      "loss": 0.0032,
+      "step": 13160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001352811513947521,
+      "loss": 0.0019,
+      "step": 13161
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013526185970339294,
+      "loss": 0.0082,
+      "step": 13162
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013524256825818854,
+      "loss": 0.0334,
+      "step": 13163
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013522327705946106,
+      "loss": 0.0214,
+      "step": 13164
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013520398610753281,
+      "loss": 0.0029,
+      "step": 13165
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013518469540272588,
+      "loss": 0.0038,
+      "step": 13166
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013516540494536253,
+      "loss": 0.0019,
+      "step": 13167
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013514611473576494,
+      "loss": 0.0128,
+      "step": 13168
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001351268247742552,
+      "loss": 0.0066,
+      "step": 13169
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001351075350611556,
+      "loss": 0.0111,
+      "step": 13170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001350882455967883,
+      "loss": 0.0158,
+      "step": 13171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013506895638147539,
+      "loss": 0.0166,
+      "step": 13172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013504966741553914,
+      "loss": 0.0117,
+      "step": 13173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001350303786993016,
+      "loss": 0.009,
+      "step": 13174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001350110902330851,
+      "loss": 0.0023,
+      "step": 13175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013499180201721167,
+      "loss": 0.0066,
+      "step": 13176
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013497251405200338,
+      "loss": 0.0053,
+      "step": 13177
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013495322633778255,
+      "loss": 0.0123,
+      "step": 13178
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001349339388748712,
+      "loss": 0.0013,
+      "step": 13179
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013491465166359155,
+      "loss": 0.0231,
+      "step": 13180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013489536470426568,
+      "loss": 0.0034,
+      "step": 13181
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013487607799721567,
+      "loss": 0.0058,
+      "step": 13182
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001348567915427638,
+      "loss": 0.0073,
+      "step": 13183
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000134837505341232,
+      "loss": 0.0082,
+      "step": 13184
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013481821939294257,
+      "loss": 0.0089,
+      "step": 13185
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013479893369821753,
+      "loss": 0.0048,
+      "step": 13186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013477964825737895,
+      "loss": 0.005,
+      "step": 13187
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013476036307074902,
+      "loss": 0.0071,
+      "step": 13188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001347410781386497,
+      "loss": 0.0114,
+      "step": 13189
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001347217934614033,
+      "loss": 0.0067,
+      "step": 13190
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001347025090393317,
+      "loss": 0.0054,
+      "step": 13191
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001346832248727571,
+      "loss": 0.0017,
+      "step": 13192
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013466394096200158,
+      "loss": 0.0072,
+      "step": 13193
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013464465730738718,
+      "loss": 0.0119,
+      "step": 13194
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013462537390923596,
+      "loss": 0.0042,
+      "step": 13195
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013460609076787009,
+      "loss": 0.0118,
+      "step": 13196
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013458680788361145,
+      "loss": 0.0101,
+      "step": 13197
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001345675252567823,
+      "loss": 0.0073,
+      "step": 13198
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013454824288770462,
+      "loss": 0.0116,
+      "step": 13199
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013452896077670038,
+      "loss": 0.0094,
+      "step": 13200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013450967892409175,
+      "loss": 0.0073,
+      "step": 13201
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001344903973302007,
+      "loss": 0.0132,
+      "step": 13202
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001344711159953493,
+      "loss": 0.0101,
+      "step": 13203
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001344518349198596,
+      "loss": 0.0013,
+      "step": 13204
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013443255410405355,
+      "loss": 0.0048,
+      "step": 13205
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013441327354825328,
+      "loss": 0.0073,
+      "step": 13206
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001343939932527807,
+      "loss": 0.0157,
+      "step": 13207
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013437471321795798,
+      "loss": 0.0025,
+      "step": 13208
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000134355433444107,
+      "loss": 0.0094,
+      "step": 13209
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001343361539315498,
+      "loss": 0.0182,
+      "step": 13210
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013431687468060845,
+      "loss": 0.011,
+      "step": 13211
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001342975956916048,
+      "loss": 0.0103,
+      "step": 13212
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000134278316964861,
+      "loss": 0.0075,
+      "step": 13213
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013425903850069904,
+      "loss": 0.0006,
+      "step": 13214
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013423976029944073,
+      "loss": 0.021,
+      "step": 13215
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013422048236140827,
+      "loss": 0.0187,
+      "step": 13216
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013420120468692355,
+      "loss": 0.0096,
+      "step": 13217
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013418192727630845,
+      "loss": 0.003,
+      "step": 13218
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001341626501298851,
+      "loss": 0.0049,
+      "step": 13219
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001341433732479753,
+      "loss": 0.0075,
+      "step": 13220
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013412409663090118,
+      "loss": 0.0019,
+      "step": 13221
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001341048202789846,
+      "loss": 0.0173,
+      "step": 13222
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013408554419254746,
+      "loss": 0.0087,
+      "step": 13223
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013406626837191184,
+      "loss": 0.0103,
+      "step": 13224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013404699281739958,
+      "loss": 0.0074,
+      "step": 13225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013402771752933269,
+      "loss": 0.0201,
+      "step": 13226
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001340084425080331,
+      "loss": 0.0074,
+      "step": 13227
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001339891677538226,
+      "loss": 0.0105,
+      "step": 13228
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001339698932670233,
+      "loss": 0.0075,
+      "step": 13229
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000133950619047957,
+      "loss": 0.0033,
+      "step": 13230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013393134509694568,
+      "loss": 0.0072,
+      "step": 13231
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013391207141431123,
+      "loss": 0.0134,
+      "step": 13232
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013389279800037553,
+      "loss": 0.0126,
+      "step": 13233
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013387352485546054,
+      "loss": 0.0106,
+      "step": 13234
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013385425197988806,
+      "loss": 0.0115,
+      "step": 13235
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013383497937398015,
+      "loss": 0.0011,
+      "step": 13236
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013381570703805857,
+      "loss": 0.0142,
+      "step": 13237
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013379643497244518,
+      "loss": 0.0054,
+      "step": 13238
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013377716317746196,
+      "loss": 0.0136,
+      "step": 13239
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001337578916534307,
+      "loss": 0.0213,
+      "step": 13240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013373862040067335,
+      "loss": 0.0005,
+      "step": 13241
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013371934941951172,
+      "loss": 0.0009,
+      "step": 13242
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013370007871026768,
+      "loss": 0.0012,
+      "step": 13243
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013368080827326308,
+      "loss": 0.0013,
+      "step": 13244
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013366153810881985,
+      "loss": 0.0069,
+      "step": 13245
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013364226821725966,
+      "loss": 0.007,
+      "step": 13246
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001336229985989046,
+      "loss": 0.0072,
+      "step": 13247
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013360372925407625,
+      "loss": 0.0111,
+      "step": 13248
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013358446018309667,
+      "loss": 0.006,
+      "step": 13249
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013356519138628758,
+      "loss": 0.0011,
+      "step": 13250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013354592286397076,
+      "loss": 0.0018,
+      "step": 13251
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013352665461646815,
+      "loss": 0.0008,
+      "step": 13252
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013350738664410144,
+      "loss": 0.013,
+      "step": 13253
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013348811894719253,
+      "loss": 0.0156,
+      "step": 13254
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013346885152606323,
+      "loss": 0.0075,
+      "step": 13255
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013344958438103526,
+      "loss": 0.0025,
+      "step": 13256
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013343031751243055,
+      "loss": 0.0092,
+      "step": 13257
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013341105092057074,
+      "loss": 0.0126,
+      "step": 13258
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013339178460577774,
+      "loss": 0.0096,
+      "step": 13259
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001333725185683733,
+      "loss": 0.002,
+      "step": 13260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013335325280867914,
+      "loss": 0.0039,
+      "step": 13261
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013333398732701712,
+      "loss": 0.0096,
+      "step": 13262
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001333147221237089,
+      "loss": 0.0072,
+      "step": 13263
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001332954571990764,
+      "loss": 0.017,
+      "step": 13264
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013327619255344128,
+      "loss": 0.0038,
+      "step": 13265
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013325692818712525,
+      "loss": 0.0123,
+      "step": 13266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001332376641004502,
+      "loss": 0.0127,
+      "step": 13267
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001332184002937378,
+      "loss": 0.0255,
+      "step": 13268
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013319913676730974,
+      "loss": 0.0218,
+      "step": 13269
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013317987352148787,
+      "loss": 0.0052,
+      "step": 13270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001331606105565938,
+      "loss": 0.003,
+      "step": 13271
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013314134787294934,
+      "loss": 0.0062,
+      "step": 13272
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001331220854708762,
+      "loss": 0.0032,
+      "step": 13273
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001331028233506961,
+      "loss": 0.0068,
+      "step": 13274
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013308356151273075,
+      "loss": 0.0077,
+      "step": 13275
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013306429995730177,
+      "loss": 0.0371,
+      "step": 13276
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013304503868473108,
+      "loss": 0.0102,
+      "step": 13277
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001330257776953402,
+      "loss": 0.0039,
+      "step": 13278
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013300651698945084,
+      "loss": 0.0062,
+      "step": 13279
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013298725656738476,
+      "loss": 0.0113,
+      "step": 13280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013296799642946357,
+      "loss": 0.009,
+      "step": 13281
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013294873657600904,
+      "loss": 0.0078,
+      "step": 13282
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001329294770073428,
+      "loss": 0.0014,
+      "step": 13283
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013291021772378648,
+      "loss": 0.0088,
+      "step": 13284
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013289095872566178,
+      "loss": 0.0102,
+      "step": 13285
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013287170001329037,
+      "loss": 0.0146,
+      "step": 13286
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001328524415869939,
+      "loss": 0.0042,
+      "step": 13287
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013283318344709406,
+      "loss": 0.0198,
+      "step": 13288
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001328139255939124,
+      "loss": 0.0051,
+      "step": 13289
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013279466802777068,
+      "loss": 0.0028,
+      "step": 13290
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001327754107489905,
+      "loss": 0.0053,
+      "step": 13291
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001327561537578934,
+      "loss": 0.0088,
+      "step": 13292
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013273689705480113,
+      "loss": 0.014,
+      "step": 13293
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013271764064003524,
+      "loss": 0.0121,
+      "step": 13294
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013269838451391737,
+      "loss": 0.0053,
+      "step": 13295
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013267912867676919,
+      "loss": 0.0113,
+      "step": 13296
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013265987312891217,
+      "loss": 0.0088,
+      "step": 13297
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013264061787066807,
+      "loss": 0.0164,
+      "step": 13298
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013262136290235837,
+      "loss": 0.0079,
+      "step": 13299
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013260210822430476,
+      "loss": 0.0082,
+      "step": 13300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001325828538368288,
+      "loss": 0.0135,
+      "step": 13301
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000132563599740252,
+      "loss": 0.0106,
+      "step": 13302
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013254434593489605,
+      "loss": 0.0018,
+      "step": 13303
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001325250924210824,
+      "loss": 0.0064,
+      "step": 13304
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013250583919913278,
+      "loss": 0.0108,
+      "step": 13305
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013248658626936863,
+      "loss": 0.0046,
+      "step": 13306
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001324673336321115,
+      "loss": 0.0052,
+      "step": 13307
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013244808128768306,
+      "loss": 0.0094,
+      "step": 13308
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013242882923640475,
+      "loss": 0.033,
+      "step": 13309
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013240957747859822,
+      "loss": 0.0116,
+      "step": 13310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013239032601458493,
+      "loss": 0.0034,
+      "step": 13311
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013237107484468638,
+      "loss": 0.0056,
+      "step": 13312
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001323518239692242,
+      "loss": 0.026,
+      "step": 13313
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013233257338851986,
+      "loss": 0.0052,
+      "step": 13314
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013231332310289492,
+      "loss": 0.0025,
+      "step": 13315
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013229407311267087,
+      "loss": 0.0044,
+      "step": 13316
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001322748234181692,
+      "loss": 0.0022,
+      "step": 13317
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013225557401971142,
+      "loss": 0.024,
+      "step": 13318
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001322363249176191,
+      "loss": 0.005,
+      "step": 13319
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013221707611221363,
+      "loss": 0.0013,
+      "step": 13320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001321978276038166,
+      "loss": 0.0183,
+      "step": 13321
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013217857939274943,
+      "loss": 0.0022,
+      "step": 13322
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013215933147933364,
+      "loss": 0.0193,
+      "step": 13323
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001321400838638907,
+      "loss": 0.0052,
+      "step": 13324
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013212083654674204,
+      "loss": 0.0054,
+      "step": 13325
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013210158952820916,
+      "loss": 0.008,
+      "step": 13326
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013208234280861353,
+      "loss": 0.0043,
+      "step": 13327
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001320630963882766,
+      "loss": 0.0028,
+      "step": 13328
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013204385026751984,
+      "loss": 0.0149,
+      "step": 13329
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001320246044466646,
+      "loss": 0.0089,
+      "step": 13330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001320053589260325,
+      "loss": 0.0045,
+      "step": 13331
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013198611370594479,
+      "loss": 0.0063,
+      "step": 13332
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013196686878672305,
+      "loss": 0.006,
+      "step": 13333
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013194762416868862,
+      "loss": 0.0218,
+      "step": 13334
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001319283798521629,
+      "loss": 0.011,
+      "step": 13335
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013190913583746743,
+      "loss": 0.008,
+      "step": 13336
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013188989212492344,
+      "loss": 0.0246,
+      "step": 13337
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001318706487148525,
+      "loss": 0.0055,
+      "step": 13338
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013185140560757595,
+      "loss": 0.0025,
+      "step": 13339
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013183216280341514,
+      "loss": 0.0042,
+      "step": 13340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013181292030269156,
+      "loss": 0.0037,
+      "step": 13341
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013179367810572655,
+      "loss": 0.0086,
+      "step": 13342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013177443621284143,
+      "loss": 0.0014,
+      "step": 13343
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013175519462435767,
+      "loss": 0.0086,
+      "step": 13344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013173595334059653,
+      "loss": 0.0091,
+      "step": 13345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013171671236187954,
+      "loss": 0.0004,
+      "step": 13346
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001316974716885279,
+      "loss": 0.0009,
+      "step": 13347
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013167823132086303,
+      "loss": 0.0026,
+      "step": 13348
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001316589912592063,
+      "loss": 0.0153,
+      "step": 13349
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013163975150387903,
+      "loss": 0.0054,
+      "step": 13350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001316205120552026,
+      "loss": 0.0069,
+      "step": 13351
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013160127291349832,
+      "loss": 0.0031,
+      "step": 13352
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013158203407908744,
+      "loss": 0.0113,
+      "step": 13353
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013156279555229144,
+      "loss": 0.0079,
+      "step": 13354
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013154355733343146,
+      "loss": 0.0051,
+      "step": 13355
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000131524319422829,
+      "loss": 0.0122,
+      "step": 13356
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013150508182080525,
+      "loss": 0.0002,
+      "step": 13357
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013148584452768154,
+      "loss": 0.0053,
+      "step": 13358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013146660754377918,
+      "loss": 0.0189,
+      "step": 13359
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001314473708694194,
+      "loss": 0.0105,
+      "step": 13360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013142813450492364,
+      "loss": 0.0019,
+      "step": 13361
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013140889845061307,
+      "loss": 0.0038,
+      "step": 13362
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013138966270680896,
+      "loss": 0.0072,
+      "step": 13363
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013137042727383265,
+      "loss": 0.0115,
+      "step": 13364
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013135119215200536,
+      "loss": 0.0128,
+      "step": 13365
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013133195734164833,
+      "loss": 0.0014,
+      "step": 13366
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013131272284308285,
+      "loss": 0.0004,
+      "step": 13367
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001312934886566302,
+      "loss": 0.0047,
+      "step": 13368
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013127425478261158,
+      "loss": 0.0007,
+      "step": 13369
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001312550212213483,
+      "loss": 0.0288,
+      "step": 13370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013123578797316146,
+      "loss": 0.0072,
+      "step": 13371
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013121655503837247,
+      "loss": 0.013,
+      "step": 13372
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001311973224173024,
+      "loss": 0.0031,
+      "step": 13373
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013117809011027262,
+      "loss": 0.0013,
+      "step": 13374
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013115885811760422,
+      "loss": 0.0067,
+      "step": 13375
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013113962643961842,
+      "loss": 0.0098,
+      "step": 13376
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013112039507663653,
+      "loss": 0.0003,
+      "step": 13377
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013110116402897958,
+      "loss": 0.0269,
+      "step": 13378
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013108193329696894,
+      "loss": 0.0175,
+      "step": 13379
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013106270288092573,
+      "loss": 0.0071,
+      "step": 13380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013104347278117108,
+      "loss": 0.003,
+      "step": 13381
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013102424299802628,
+      "loss": 0.0022,
+      "step": 13382
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013100501353181235,
+      "loss": 0.0052,
+      "step": 13383
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013098578438285063,
+      "loss": 0.0018,
+      "step": 13384
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013096655555146222,
+      "loss": 0.0295,
+      "step": 13385
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013094732703796818,
+      "loss": 0.0175,
+      "step": 13386
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001309280988426898,
+      "loss": 0.0082,
+      "step": 13387
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013090887096594813,
+      "loss": 0.0141,
+      "step": 13388
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001308896434080644,
+      "loss": 0.0063,
+      "step": 13389
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013087041616935966,
+      "loss": 0.0007,
+      "step": 13390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013085118925015504,
+      "loss": 0.0258,
+      "step": 13391
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001308319626507718,
+      "loss": 0.0033,
+      "step": 13392
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013081273637153095,
+      "loss": 0.0162,
+      "step": 13393
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013079351041275355,
+      "loss": 0.0198,
+      "step": 13394
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013077428477476088,
+      "loss": 0.032,
+      "step": 13395
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013075505945787382,
+      "loss": 0.0081,
+      "step": 13396
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001307358344624137,
+      "loss": 0.0056,
+      "step": 13397
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013071660978870148,
+      "loss": 0.0081,
+      "step": 13398
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013069738543705826,
+      "loss": 0.0082,
+      "step": 13399
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013067816140780516,
+      "loss": 0.0017,
+      "step": 13400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013065893770126318,
+      "loss": 0.0038,
+      "step": 13401
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013063971431775354,
+      "loss": 0.0044,
+      "step": 13402
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001306204912575972,
+      "loss": 0.0054,
+      "step": 13403
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013060126852111517,
+      "loss": 0.0221,
+      "step": 13404
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013058204610862862,
+      "loss": 0.0098,
+      "step": 13405
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001305628240204585,
+      "loss": 0.0055,
+      "step": 13406
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000130543602256926,
+      "loss": 0.0052,
+      "step": 13407
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013052438081835203,
+      "loss": 0.0064,
+      "step": 13408
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013050515970505766,
+      "loss": 0.0111,
+      "step": 13409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001304859389173639,
+      "loss": 0.0009,
+      "step": 13410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013046671845559182,
+      "loss": 0.0057,
+      "step": 13411
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001304474983200624,
+      "loss": 0.0072,
+      "step": 13412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001304282785110967,
+      "loss": 0.0047,
+      "step": 13413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013040905902901562,
+      "loss": 0.0064,
+      "step": 13414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013038983987414033,
+      "loss": 0.0071,
+      "step": 13415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013037062104679173,
+      "loss": 0.0022,
+      "step": 13416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001303514025472907,
+      "loss": 0.0102,
+      "step": 13417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013033218437595844,
+      "loss": 0.0024,
+      "step": 13418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013031296653311574,
+      "loss": 0.009,
+      "step": 13419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001302937490190837,
+      "loss": 0.0131,
+      "step": 13420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001302745318341833,
+      "loss": 0.006,
+      "step": 13421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001302553149787354,
+      "loss": 0.0142,
+      "step": 13422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013023609845306102,
+      "loss": 0.0278,
+      "step": 13423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013021688225748108,
+      "loss": 0.0062,
+      "step": 13424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001301976663923166,
+      "loss": 0.0129,
+      "step": 13425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013017845085788846,
+      "loss": 0.0078,
+      "step": 13426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013015923565451757,
+      "loss": 0.0082,
+      "step": 13427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013014002078252495,
+      "loss": 0.0039,
+      "step": 13428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013012080624223138,
+      "loss": 0.0054,
+      "step": 13429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013010159203395797,
+      "loss": 0.0109,
+      "step": 13430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001300823781580255,
+      "loss": 0.0036,
+      "step": 13431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00013006316461475486,
+      "loss": 0.0051,
+      "step": 13432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00013004395140446708,
+      "loss": 0.0047,
+      "step": 13433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00013002473852748292,
+      "loss": 0.0073,
+      "step": 13434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00013000552598412338,
+      "loss": 0.0045,
+      "step": 13435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012998631377470933,
+      "loss": 0.0048,
+      "step": 13436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012996710189956153,
+      "loss": 0.0061,
+      "step": 13437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000129947890359001,
+      "loss": 0.0048,
+      "step": 13438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012992867915334852,
+      "loss": 0.0047,
+      "step": 13439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012990946828292502,
+      "loss": 0.0278,
+      "step": 13440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001298902577480513,
+      "loss": 0.008,
+      "step": 13441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012987104754904824,
+      "loss": 0.0038,
+      "step": 13442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012985183768623668,
+      "loss": 0.0101,
+      "step": 13443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001298326281599375,
+      "loss": 0.0058,
+      "step": 13444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001298134189704714,
+      "loss": 0.0116,
+      "step": 13445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012979421011815941,
+      "loss": 0.0175,
+      "step": 13446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012977500160332218,
+      "loss": 0.0103,
+      "step": 13447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012975579342628066,
+      "loss": 0.0136,
+      "step": 13448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001297365855873556,
+      "loss": 0.0031,
+      "step": 13449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012971737808686775,
+      "loss": 0.0045,
+      "step": 13450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012969817092513803,
+      "loss": 0.0154,
+      "step": 13451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012967896410248715,
+      "loss": 0.0055,
+      "step": 13452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012965975761923592,
+      "loss": 0.0014,
+      "step": 13453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012964055147570517,
+      "loss": 0.0018,
+      "step": 13454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012962134567221558,
+      "loss": 0.0123,
+      "step": 13455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012960214020908808,
+      "loss": 0.0053,
+      "step": 13456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012958293508664323,
+      "loss": 0.0005,
+      "step": 13457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012956373030520202,
+      "loss": 0.0059,
+      "step": 13458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012954452586508506,
+      "loss": 0.0084,
+      "step": 13459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012952532176661305,
+      "loss": 0.0109,
+      "step": 13460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012950611801010686,
+      "loss": 0.0067,
+      "step": 13461
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012948691459588717,
+      "loss": 0.0039,
+      "step": 13462
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012946771152427477,
+      "loss": 0.0128,
+      "step": 13463
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012944850879559035,
+      "loss": 0.0052,
+      "step": 13464
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012942930641015457,
+      "loss": 0.0067,
+      "step": 13465
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012941010436828826,
+      "loss": 0.0087,
+      "step": 13466
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012939090267031207,
+      "loss": 0.002,
+      "step": 13467
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012937170131654664,
+      "loss": 0.0072,
+      "step": 13468
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012935250030731284,
+      "loss": 0.0042,
+      "step": 13469
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012933329964293113,
+      "loss": 0.0102,
+      "step": 13470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012931409932372245,
+      "loss": 0.0044,
+      "step": 13471
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001292948993500073,
+      "loss": 0.031,
+      "step": 13472
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012927569972210639,
+      "loss": 0.001,
+      "step": 13473
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012925650044034046,
+      "loss": 0.0031,
+      "step": 13474
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012923730150503007,
+      "loss": 0.0021,
+      "step": 13475
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012921810291649603,
+      "loss": 0.0077,
+      "step": 13476
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012919890467505887,
+      "loss": 0.0051,
+      "step": 13477
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001291797067810392,
+      "loss": 0.0087,
+      "step": 13478
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001291605092347578,
+      "loss": 0.0136,
+      "step": 13479
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001291413120365352,
+      "loss": 0.0119,
+      "step": 13480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001291221151866921,
+      "loss": 0.0062,
+      "step": 13481
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012910291868554905,
+      "loss": 0.0017,
+      "step": 13482
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001290837225334267,
+      "loss": 0.0054,
+      "step": 13483
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001290645267306457,
+      "loss": 0.0102,
+      "step": 13484
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012904533127752654,
+      "loss": 0.0172,
+      "step": 13485
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012902613617439,
+      "loss": 0.0073,
+      "step": 13486
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012900694142155657,
+      "loss": 0.0102,
+      "step": 13487
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012898774701934677,
+      "loss": 0.0017,
+      "step": 13488
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012896855296808135,
+      "loss": 0.0048,
+      "step": 13489
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012894935926808075,
+      "loss": 0.0034,
+      "step": 13490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012893016591966554,
+      "loss": 0.0014,
+      "step": 13491
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012891097292315638,
+      "loss": 0.008,
+      "step": 13492
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012889178027887378,
+      "loss": 0.0034,
+      "step": 13493
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001288725879871383,
+      "loss": 0.0029,
+      "step": 13494
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001288533960482705,
+      "loss": 0.0135,
+      "step": 13495
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012883420446259083,
+      "loss": 0.0059,
+      "step": 13496
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012881501323041997,
+      "loss": 0.0048,
+      "step": 13497
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012879582235207832,
+      "loss": 0.0031,
+      "step": 13498
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012877663182788652,
+      "loss": 0.0154,
+      "step": 13499
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012875744165816502,
+      "loss": 0.0032,
+      "step": 13500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001287382518432343,
+      "loss": 0.004,
+      "step": 13501
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012871906238341496,
+      "loss": 0.0065,
+      "step": 13502
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001286998732790274,
+      "loss": 0.0087,
+      "step": 13503
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012868068453039216,
+      "loss": 0.0007,
+      "step": 13504
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012866149613782981,
+      "loss": 0.0046,
+      "step": 13505
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012864230810166069,
+      "loss": 0.0134,
+      "step": 13506
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012862312042220535,
+      "loss": 0.0174,
+      "step": 13507
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012860393309978424,
+      "loss": 0.0146,
+      "step": 13508
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012858474613471787,
+      "loss": 0.0069,
+      "step": 13509
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012856555952732668,
+      "loss": 0.0238,
+      "step": 13510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012854637327793102,
+      "loss": 0.0193,
+      "step": 13511
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001285271873868515,
+      "loss": 0.0043,
+      "step": 13512
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012850800185440842,
+      "loss": 0.0216,
+      "step": 13513
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284888166809223,
+      "loss": 0.0062,
+      "step": 13514
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284696318667136,
+      "loss": 0.006,
+      "step": 13515
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012845044741210262,
+      "loss": 0.0051,
+      "step": 13516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284312633174099,
+      "loss": 0.0096,
+      "step": 13517
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012841207958295582,
+      "loss": 0.0147,
+      "step": 13518
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012839289620906064,
+      "loss": 0.0114,
+      "step": 13519
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012837371319604499,
+      "loss": 0.0111,
+      "step": 13520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012835453054422908,
+      "loss": 0.0007,
+      "step": 13521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012833534825393343,
+      "loss": 0.0076,
+      "step": 13522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012831616632547833,
+      "loss": 0.0052,
+      "step": 13523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012829698475918418,
+      "loss": 0.0116,
+      "step": 13524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012827780355537135,
+      "loss": 0.0384,
+      "step": 13525
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012825862271436018,
+      "loss": 0.0037,
+      "step": 13526
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012823944223647112,
+      "loss": 0.0036,
+      "step": 13527
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012822026212202443,
+      "loss": 0.0076,
+      "step": 13528
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012820108237134042,
+      "loss": 0.0008,
+      "step": 13529
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012818190298473955,
+      "loss": 0.0151,
+      "step": 13530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000128162723962542,
+      "loss": 0.0096,
+      "step": 13531
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001281435453050683,
+      "loss": 0.0094,
+      "step": 13532
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012812436701263856,
+      "loss": 0.0085,
+      "step": 13533
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012810518908557323,
+      "loss": 0.022,
+      "step": 13534
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012808601152419254,
+      "loss": 0.0014,
+      "step": 13535
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012806683432881684,
+      "loss": 0.0089,
+      "step": 13536
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012804765749976639,
+      "loss": 0.0068,
+      "step": 13537
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012802848103736153,
+      "loss": 0.0078,
+      "step": 13538
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012800930494192246,
+      "loss": 0.0014,
+      "step": 13539
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012799012921376957,
+      "loss": 0.0039,
+      "step": 13540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012797095385322306,
+      "loss": 0.0077,
+      "step": 13541
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012795177886060316,
+      "loss": 0.003,
+      "step": 13542
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012793260423623023,
+      "loss": 0.0155,
+      "step": 13543
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012791342998042437,
+      "loss": 0.0213,
+      "step": 13544
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000127894256093506,
+      "loss": 0.0269,
+      "step": 13545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012787508257579527,
+      "loss": 0.002,
+      "step": 13546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001278559094276124,
+      "loss": 0.0097,
+      "step": 13547
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012783673664927767,
+      "loss": 0.0007,
+      "step": 13548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001278175642411112,
+      "loss": 0.0042,
+      "step": 13549
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001277983922034334,
+      "loss": 0.0274,
+      "step": 13550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012777922053656434,
+      "loss": 0.0077,
+      "step": 13551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012776004924082413,
+      "loss": 0.0126,
+      "step": 13552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012774087831653319,
+      "loss": 0.023,
+      "step": 13553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001277217077640115,
+      "loss": 0.0086,
+      "step": 13554
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012770253758357945,
+      "loss": 0.0016,
+      "step": 13555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012768336777555707,
+      "loss": 0.0136,
+      "step": 13556
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012766419834026453,
+      "loss": 0.0067,
+      "step": 13557
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001276450292780221,
+      "loss": 0.0202,
+      "step": 13558
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012762586058914983,
+      "loss": 0.0257,
+      "step": 13559
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000127606692273968,
+      "loss": 0.0043,
+      "step": 13560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012758752433279663,
+      "loss": 0.0105,
+      "step": 13561
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012756835676595588,
+      "loss": 0.008,
+      "step": 13562
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012754918957376597,
+      "loss": 0.0051,
+      "step": 13563
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012753002275654694,
+      "loss": 0.0025,
+      "step": 13564
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012751085631461894,
+      "loss": 0.0016,
+      "step": 13565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001274916902483021,
+      "loss": 0.005,
+      "step": 13566
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001274725245579165,
+      "loss": 0.0049,
+      "step": 13567
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001274533592437823,
+      "loss": 0.0025,
+      "step": 13568
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012743419430621958,
+      "loss": 0.0011,
+      "step": 13569
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012741502974554834,
+      "loss": 0.0108,
+      "step": 13570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001273958655620888,
+      "loss": 0.0022,
+      "step": 13571
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001273767017561609,
+      "loss": 0.005,
+      "step": 13572
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012735753832808486,
+      "loss": 0.0008,
+      "step": 13573
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012733837527818066,
+      "loss": 0.0119,
+      "step": 13574
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012731921260676835,
+      "loss": 0.0046,
+      "step": 13575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012730005031416798,
+      "loss": 0.0045,
+      "step": 13576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012728088840069965,
+      "loss": 0.0065,
+      "step": 13577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012726172686668334,
+      "loss": 0.011,
+      "step": 13578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012724256571243917,
+      "loss": 0.0047,
+      "step": 13579
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012722340493828704,
+      "loss": 0.0009,
+      "step": 13580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001272042445445471,
+      "loss": 0.0088,
+      "step": 13581
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012718508453153925,
+      "loss": 0.0217,
+      "step": 13582
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012716592489958362,
+      "loss": 0.0062,
+      "step": 13583
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012714676564900014,
+      "loss": 0.0165,
+      "step": 13584
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012712760678010876,
+      "loss": 0.0122,
+      "step": 13585
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012710844829322954,
+      "loss": 0.007,
+      "step": 13586
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012708929018868245,
+      "loss": 0.007,
+      "step": 13587
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012707013246678745,
+      "loss": 0.0067,
+      "step": 13588
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012705097512786458,
+      "loss": 0.006,
+      "step": 13589
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012703181817223365,
+      "loss": 0.0035,
+      "step": 13590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001270126616002148,
+      "loss": 0.0077,
+      "step": 13591
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001269935054121279,
+      "loss": 0.006,
+      "step": 13592
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012697434960829278,
+      "loss": 0.0127,
+      "step": 13593
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001269551941890296,
+      "loss": 0.0141,
+      "step": 13594
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012693603915465805,
+      "loss": 0.0078,
+      "step": 13595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001269168845054983,
+      "loss": 0.0062,
+      "step": 13596
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001268977302418701,
+      "loss": 0.0358,
+      "step": 13597
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001268785763640934,
+      "loss": 0.001,
+      "step": 13598
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012685942287248816,
+      "loss": 0.0032,
+      "step": 13599
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012684026976737418,
+      "loss": 0.0141,
+      "step": 13600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001268211170490715,
+      "loss": 0.0128,
+      "step": 13601
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001268019647178999,
+      "loss": 0.0203,
+      "step": 13602
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012678281277417918,
+      "loss": 0.0043,
+      "step": 13603
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012676366121822943,
+      "loss": 0.0263,
+      "step": 13604
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001267445100503703,
+      "loss": 0.0072,
+      "step": 13605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012672535927092183,
+      "loss": 0.0131,
+      "step": 13606
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012670620888020377,
+      "loss": 0.0079,
+      "step": 13607
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012668705887853597,
+      "loss": 0.0021,
+      "step": 13608
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001266679092662383,
+      "loss": 0.0091,
+      "step": 13609
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012664876004363058,
+      "loss": 0.0148,
+      "step": 13610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012662961121103267,
+      "loss": 0.0052,
+      "step": 13611
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001266104627687644,
+      "loss": 0.0024,
+      "step": 13612
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012659131471714547,
+      "loss": 0.0048,
+      "step": 13613
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012657216705649584,
+      "loss": 0.0055,
+      "step": 13614
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012655301978713526,
+      "loss": 0.0143,
+      "step": 13615
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012653387290938344,
+      "loss": 0.0116,
+      "step": 13616
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012651472642356024,
+      "loss": 0.0115,
+      "step": 13617
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012649558032998547,
+      "loss": 0.0083,
+      "step": 13618
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012647643462897887,
+      "loss": 0.0052,
+      "step": 13619
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012645728932086022,
+      "loss": 0.0021,
+      "step": 13620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012643814440594926,
+      "loss": 0.0248,
+      "step": 13621
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001264189998845658,
+      "loss": 0.0011,
+      "step": 13622
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001263998557570295,
+      "loss": 0.0278,
+      "step": 13623
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001263807120236602,
+      "loss": 0.0019,
+      "step": 13624
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012636156868477765,
+      "loss": 0.0036,
+      "step": 13625
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001263424257407014,
+      "loss": 0.0084,
+      "step": 13626
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001263232831917514,
+      "loss": 0.0096,
+      "step": 13627
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012630414103824718,
+      "loss": 0.0113,
+      "step": 13628
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012628499928050857,
+      "loss": 0.0043,
+      "step": 13629
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012626585791885526,
+      "loss": 0.0045,
+      "step": 13630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012624671695360686,
+      "loss": 0.0041,
+      "step": 13631
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001262275763850832,
+      "loss": 0.0115,
+      "step": 13632
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012620843621360383,
+      "loss": 0.0261,
+      "step": 13633
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012618929643948854,
+      "loss": 0.0043,
+      "step": 13634
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012617015706305694,
+      "loss": 0.0073,
+      "step": 13635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012615101808462865,
+      "loss": 0.0105,
+      "step": 13636
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001261318795045234,
+      "loss": 0.0052,
+      "step": 13637
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012611274132306083,
+      "loss": 0.004,
+      "step": 13638
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012609360354056055,
+      "loss": 0.0014,
+      "step": 13639
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012607446615734225,
+      "loss": 0.008,
+      "step": 13640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012605532917372547,
+      "loss": 0.0094,
+      "step": 13641
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012603619259002994,
+      "loss": 0.004,
+      "step": 13642
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012601705640657523,
+      "loss": 0.008,
+      "step": 13643
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001259979206236809,
+      "loss": 0.0072,
+      "step": 13644
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012597878524166667,
+      "loss": 0.0049,
+      "step": 13645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000125959650260852,
+      "loss": 0.0029,
+      "step": 13646
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012594051568155662,
+      "loss": 0.003,
+      "step": 13647
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001259213815041,
+      "loss": 0.017,
+      "step": 13648
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012590224772880178,
+      "loss": 0.0055,
+      "step": 13649
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012588311435598147,
+      "loss": 0.0029,
+      "step": 13650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012586398138595867,
+      "loss": 0.0069,
+      "step": 13651
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012584484881905303,
+      "loss": 0.0045,
+      "step": 13652
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012582571665558395,
+      "loss": 0.0055,
+      "step": 13653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012580658489587103,
+      "loss": 0.0104,
+      "step": 13654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012578745354023383,
+      "loss": 0.0066,
+      "step": 13655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012576832258899182,
+      "loss": 0.0122,
+      "step": 13656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012574919204246465,
+      "loss": 0.0061,
+      "step": 13657
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012573006190097168,
+      "loss": 0.0043,
+      "step": 13658
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012571093216483248,
+      "loss": 0.0056,
+      "step": 13659
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001256918028343666,
+      "loss": 0.0188,
+      "step": 13660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012567267390989348,
+      "loss": 0.0006,
+      "step": 13661
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012565354539173263,
+      "loss": 0.0466,
+      "step": 13662
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012563441728020354,
+      "loss": 0.0147,
+      "step": 13663
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001256152895756256,
+      "loss": 0.0065,
+      "step": 13664
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012559616227831848,
+      "loss": 0.0058,
+      "step": 13665
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012557703538860148,
+      "loss": 0.0037,
+      "step": 13666
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012555790890679403,
+      "loss": 0.0076,
+      "step": 13667
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001255387828332157,
+      "loss": 0.0046,
+      "step": 13668
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001255196571681858,
+      "loss": 0.0042,
+      "step": 13669
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012550053191202385,
+      "loss": 0.004,
+      "step": 13670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012548140706504933,
+      "loss": 0.02,
+      "step": 13671
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012546228262758153,
+      "loss": 0.0148,
+      "step": 13672
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012544315859993997,
+      "loss": 0.0098,
+      "step": 13673
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012542403498244395,
+      "loss": 0.0193,
+      "step": 13674
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012540491177541303,
+      "loss": 0.0021,
+      "step": 13675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001253857889791665,
+      "loss": 0.0025,
+      "step": 13676
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012536666659402368,
+      "loss": 0.0037,
+      "step": 13677
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012534754462030411,
+      "loss": 0.0162,
+      "step": 13678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000125328423058327,
+      "loss": 0.0065,
+      "step": 13679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012530930190841186,
+      "loss": 0.01,
+      "step": 13680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012529018117087796,
+      "loss": 0.0084,
+      "step": 13681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012527106084604463,
+      "loss": 0.0108,
+      "step": 13682
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012525194093423133,
+      "loss": 0.0086,
+      "step": 13683
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012523282143575727,
+      "loss": 0.0016,
+      "step": 13684
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012521370235094191,
+      "loss": 0.008,
+      "step": 13685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012519458368010451,
+      "loss": 0.0031,
+      "step": 13686
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012517546542356432,
+      "loss": 0.0228,
+      "step": 13687
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012515634758164082,
+      "loss": 0.0163,
+      "step": 13688
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001251372301546531,
+      "loss": 0.0028,
+      "step": 13689
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012511811314292064,
+      "loss": 0.0076,
+      "step": 13690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012509899654676262,
+      "loss": 0.0037,
+      "step": 13691
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012507988036649834,
+      "loss": 0.0113,
+      "step": 13692
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012506076460244716,
+      "loss": 0.0159,
+      "step": 13693
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012504164925492827,
+      "loss": 0.0103,
+      "step": 13694
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001250225343242609,
+      "loss": 0.0013,
+      "step": 13695
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012500341981076446,
+      "loss": 0.0073,
+      "step": 13696
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012498430571475797,
+      "loss": 0.0096,
+      "step": 13697
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012496519203656088,
+      "loss": 0.0031,
+      "step": 13698
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012494607877649232,
+      "loss": 0.0044,
+      "step": 13699
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001249269659348715,
+      "loss": 0.0005,
+      "step": 13700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012490785351201774,
+      "loss": 0.0043,
+      "step": 13701
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012488874150825015,
+      "loss": 0.0094,
+      "step": 13702
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012486962992388798,
+      "loss": 0.0029,
+      "step": 13703
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012485051875925045,
+      "loss": 0.0043,
+      "step": 13704
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012483140801465667,
+      "loss": 0.0004,
+      "step": 13705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012481229769042596,
+      "loss": 0.0058,
+      "step": 13706
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012479318778687738,
+      "loss": 0.0025,
+      "step": 13707
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001247740783043302,
+      "loss": 0.0183,
+      "step": 13708
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012475496924310348,
+      "loss": 0.0118,
+      "step": 13709
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001247358606035164,
+      "loss": 0.0055,
+      "step": 13710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001247167523858882,
+      "loss": 0.0086,
+      "step": 13711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012469764459053793,
+      "loss": 0.0082,
+      "step": 13712
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012467853721778472,
+      "loss": 0.0103,
+      "step": 13713
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012465943026794782,
+      "loss": 0.0024,
+      "step": 13714
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012464032374134618,
+      "loss": 0.0054,
+      "step": 13715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012462121763829906,
+      "loss": 0.0115,
+      "step": 13716
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012460211195912553,
+      "loss": 0.0095,
+      "step": 13717
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012458300670414461,
+      "loss": 0.0041,
+      "step": 13718
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012456390187367552,
+      "loss": 0.0121,
+      "step": 13719
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012454479746803722,
+      "loss": 0.006,
+      "step": 13720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001245256934875489,
+      "loss": 0.0128,
+      "step": 13721
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012450658993252956,
+      "loss": 0.0065,
+      "step": 13722
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012448748680329825,
+      "loss": 0.0129,
+      "step": 13723
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012446838410017417,
+      "loss": 0.0169,
+      "step": 13724
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012444928182347615,
+      "loss": 0.0064,
+      "step": 13725
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012443017997352347,
+      "loss": 0.0026,
+      "step": 13726
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012441107855063502,
+      "loss": 0.0049,
+      "step": 13727
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001243919775551298,
+      "loss": 0.0042,
+      "step": 13728
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012437287698732698,
+      "loss": 0.0204,
+      "step": 13729
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001243537768475454,
+      "loss": 0.0047,
+      "step": 13730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012433467713610422,
+      "loss": 0.0072,
+      "step": 13731
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012431557785332235,
+      "loss": 0.0106,
+      "step": 13732
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012429647899951882,
+      "loss": 0.011,
+      "step": 13733
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012427738057501261,
+      "loss": 0.0079,
+      "step": 13734
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012425828258012266,
+      "loss": 0.0035,
+      "step": 13735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012423918501516806,
+      "loss": 0.004,
+      "step": 13736
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001242200878804677,
+      "loss": 0.005,
+      "step": 13737
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012420099117634045,
+      "loss": 0.0333,
+      "step": 13738
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001241818949031054,
+      "loss": 0.0059,
+      "step": 13739
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012416279906108147,
+      "loss": 0.0124,
+      "step": 13740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012414370365058748,
+      "loss": 0.0047,
+      "step": 13741
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012412460867194253,
+      "loss": 0.008,
+      "step": 13742
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001241055141254654,
+      "loss": 0.0235,
+      "step": 13743
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001240864200114751,
+      "loss": 0.0105,
+      "step": 13744
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001240673263302905,
+      "loss": 0.0189,
+      "step": 13745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012404823308223044,
+      "loss": 0.0419,
+      "step": 13746
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012402914026761396,
+      "loss": 0.0036,
+      "step": 13747
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001240100478867598,
+      "loss": 0.0094,
+      "step": 13748
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012399095593998693,
+      "loss": 0.0062,
+      "step": 13749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012397186442761422,
+      "loss": 0.0101,
+      "step": 13750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012395277334996044,
+      "loss": 0.0126,
+      "step": 13751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012393368270734455,
+      "loss": 0.0061,
+      "step": 13752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012391459250008535,
+      "loss": 0.005,
+      "step": 13753
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012389550272850168,
+      "loss": 0.0134,
+      "step": 13754
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012387641339291246,
+      "loss": 0.0011,
+      "step": 13755
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012385732449363636,
+      "loss": 0.0071,
+      "step": 13756
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012383823603099236,
+      "loss": 0.012,
+      "step": 13757
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012381914800529913,
+      "loss": 0.002,
+      "step": 13758
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001238000604168756,
+      "loss": 0.0038,
+      "step": 13759
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012378097326604055,
+      "loss": 0.0062,
+      "step": 13760
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012376188655311267,
+      "loss": 0.0047,
+      "step": 13761
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012374280027841088,
+      "loss": 0.0042,
+      "step": 13762
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012372371444225382,
+      "loss": 0.0047,
+      "step": 13763
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001237046290449603,
+      "loss": 0.0117,
+      "step": 13764
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001236855440868492,
+      "loss": 0.0029,
+      "step": 13765
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012366645956823913,
+      "loss": 0.0132,
+      "step": 13766
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012364737548944895,
+      "loss": 0.0188,
+      "step": 13767
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012362829185079733,
+      "loss": 0.0066,
+      "step": 13768
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012360920865260296,
+      "loss": 0.022,
+      "step": 13769
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012359012589518469,
+      "loss": 0.0026,
+      "step": 13770
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001235710435788611,
+      "loss": 0.0048,
+      "step": 13771
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012355196170395102,
+      "loss": 0.0064,
+      "step": 13772
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012353288027077309,
+      "loss": 0.0158,
+      "step": 13773
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012351379927964603,
+      "loss": 0.0287,
+      "step": 13774
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001234947187308885,
+      "loss": 0.0057,
+      "step": 13775
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012347563862481918,
+      "loss": 0.0095,
+      "step": 13776
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012345655896175682,
+      "loss": 0.0083,
+      "step": 13777
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012343747974202,
+      "loss": 0.0027,
+      "step": 13778
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001234184009659274,
+      "loss": 0.0101,
+      "step": 13779
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012339932263379772,
+      "loss": 0.0055,
+      "step": 13780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001233802447459495,
+      "loss": 0.0016,
+      "step": 13781
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012336116730270148,
+      "loss": 0.0044,
+      "step": 13782
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012334209030437224,
+      "loss": 0.0123,
+      "step": 13783
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012332301375128044,
+      "loss": 0.0163,
+      "step": 13784
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012330393764374464,
+      "loss": 0.0283,
+      "step": 13785
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001232848619820835,
+      "loss": 0.0137,
+      "step": 13786
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012326578676661551,
+      "loss": 0.0155,
+      "step": 13787
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012324671199765946,
+      "loss": 0.0103,
+      "step": 13788
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012322763767553372,
+      "loss": 0.0037,
+      "step": 13789
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012320856380055705,
+      "loss": 0.0033,
+      "step": 13790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012318949037304794,
+      "loss": 0.0003,
+      "step": 13791
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012317041739332487,
+      "loss": 0.008,
+      "step": 13792
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012315134486170655,
+      "loss": 0.0139,
+      "step": 13793
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012313227277851137,
+      "loss": 0.0052,
+      "step": 13794
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012311320114405798,
+      "loss": 0.0157,
+      "step": 13795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012309412995866495,
+      "loss": 0.0036,
+      "step": 13796
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012307505922265065,
+      "loss": 0.0103,
+      "step": 13797
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012305598893633374,
+      "loss": 0.003,
+      "step": 13798
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012303691910003263,
+      "loss": 0.0116,
+      "step": 13799
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012301784971406593,
+      "loss": 0.0076,
+      "step": 13800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012299878077875207,
+      "loss": 0.01,
+      "step": 13801
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012297971229440947,
+      "loss": 0.0015,
+      "step": 13802
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012296064426135672,
+      "loss": 0.0149,
+      "step": 13803
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012294157667991224,
+      "loss": 0.0015,
+      "step": 13804
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012292250955039453,
+      "loss": 0.003,
+      "step": 13805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012290344287312202,
+      "loss": 0.0088,
+      "step": 13806
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012288437664841309,
+      "loss": 0.0064,
+      "step": 13807
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012286531087658635,
+      "loss": 0.0134,
+      "step": 13808
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012284624555796006,
+      "loss": 0.0014,
+      "step": 13809
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001228271806928528,
+      "loss": 0.0117,
+      "step": 13810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001228081162815829,
+      "loss": 0.0036,
+      "step": 13811
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012278905232446875,
+      "loss": 0.0035,
+      "step": 13812
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012276998882182884,
+      "loss": 0.0092,
+      "step": 13813
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012275092577398146,
+      "loss": 0.0072,
+      "step": 13814
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001227318631812451,
+      "loss": 0.0041,
+      "step": 13815
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012271280104393805,
+      "loss": 0.0056,
+      "step": 13816
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012269373936237875,
+      "loss": 0.0098,
+      "step": 13817
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012267467813688557,
+      "loss": 0.0027,
+      "step": 13818
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012265561736777688,
+      "loss": 0.0034,
+      "step": 13819
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001226365570553709,
+      "loss": 0.0053,
+      "step": 13820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012261749719998615,
+      "loss": 0.0015,
+      "step": 13821
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012259843780194083,
+      "loss": 0.0045,
+      "step": 13822
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012257937886155337,
+      "loss": 0.0111,
+      "step": 13823
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000122560320379142,
+      "loss": 0.0045,
+      "step": 13824
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001225412623550251,
+      "loss": 0.008,
+      "step": 13825
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012252220478952094,
+      "loss": 0.0069,
+      "step": 13826
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012250314768294784,
+      "loss": 0.0012,
+      "step": 13827
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012248409103562404,
+      "loss": 0.0256,
+      "step": 13828
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012246503484786793,
+      "loss": 0.0024,
+      "step": 13829
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012244597911999764,
+      "loss": 0.0074,
+      "step": 13830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012242692385233157,
+      "loss": 0.0009,
+      "step": 13831
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012240786904518786,
+      "loss": 0.0275,
+      "step": 13832
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012238881469888487,
+      "loss": 0.0135,
+      "step": 13833
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012236976081374083,
+      "loss": 0.0077,
+      "step": 13834
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012235070739007386,
+      "loss": 0.0162,
+      "step": 13835
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012233165442820227,
+      "loss": 0.0102,
+      "step": 13836
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012231260192844433,
+      "loss": 0.0095,
+      "step": 13837
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012229354989111814,
+      "loss": 0.0132,
+      "step": 13838
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012227449831654204,
+      "loss": 0.01,
+      "step": 13839
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012225544720503409,
+      "loss": 0.0076,
+      "step": 13840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001222363965569126,
+      "loss": 0.0115,
+      "step": 13841
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001222173463724957,
+      "loss": 0.0049,
+      "step": 13842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001221982966521015,
+      "loss": 0.019,
+      "step": 13843
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012217924739604827,
+      "loss": 0.0048,
+      "step": 13844
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012216019860465405,
+      "loss": 0.0106,
+      "step": 13845
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012214115027823718,
+      "loss": 0.0042,
+      "step": 13846
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012212210241711562,
+      "loss": 0.0127,
+      "step": 13847
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012210305502160753,
+      "loss": 0.0051,
+      "step": 13848
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012208400809203117,
+      "loss": 0.009,
+      "step": 13849
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001220649616287045,
+      "loss": 0.0049,
+      "step": 13850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012204591563194575,
+      "loss": 0.0122,
+      "step": 13851
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012202687010207299,
+      "loss": 0.0113,
+      "step": 13852
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012200782503940424,
+      "loss": 0.0064,
+      "step": 13853
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001219887804442577,
+      "loss": 0.0065,
+      "step": 13854
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012196973631695135,
+      "loss": 0.0071,
+      "step": 13855
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012195069265780338,
+      "loss": 0.019,
+      "step": 13856
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012193164946713175,
+      "loss": 0.0048,
+      "step": 13857
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012191260674525455,
+      "loss": 0.0219,
+      "step": 13858
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012189356449248985,
+      "loss": 0.0039,
+      "step": 13859
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012187452270915571,
+      "loss": 0.0052,
+      "step": 13860
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012185548139557006,
+      "loss": 0.0155,
+      "step": 13861
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012183644055205107,
+      "loss": 0.0051,
+      "step": 13862
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012181740017891662,
+      "loss": 0.0041,
+      "step": 13863
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012179836027648484,
+      "loss": 0.0063,
+      "step": 13864
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012177932084507367,
+      "loss": 0.0096,
+      "step": 13865
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012176028188500107,
+      "loss": 0.0031,
+      "step": 13866
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001217412433965851,
+      "loss": 0.0041,
+      "step": 13867
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012172220538014369,
+      "loss": 0.0077,
+      "step": 13868
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012170316783599483,
+      "loss": 0.0124,
+      "step": 13869
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012168413076445654,
+      "loss": 0.0012,
+      "step": 13870
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012166509416584661,
+      "loss": 0.007,
+      "step": 13871
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001216460580404832,
+      "loss": 0.0227,
+      "step": 13872
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012162702238868405,
+      "loss": 0.0055,
+      "step": 13873
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012160798721076727,
+      "loss": 0.0143,
+      "step": 13874
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012158895250705068,
+      "loss": 0.0089,
+      "step": 13875
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012156991827785214,
+      "loss": 0.0123,
+      "step": 13876
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012155088452348968,
+      "loss": 0.0126,
+      "step": 13877
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012153185124428114,
+      "loss": 0.0035,
+      "step": 13878
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012151281844054443,
+      "loss": 0.0036,
+      "step": 13879
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012149378611259745,
+      "loss": 0.0223,
+      "step": 13880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000121474754260758,
+      "loss": 0.0039,
+      "step": 13881
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012145572288534405,
+      "loss": 0.0053,
+      "step": 13882
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012143669198667335,
+      "loss": 0.0067,
+      "step": 13883
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012141766156506387,
+      "loss": 0.0132,
+      "step": 13884
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001213986316208334,
+      "loss": 0.0119,
+      "step": 13885
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012137960215429971,
+      "loss": 0.0034,
+      "step": 13886
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012136057316578075,
+      "loss": 0.0095,
+      "step": 13887
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012134154465559424,
+      "loss": 0.0054,
+      "step": 13888
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000121322516624058,
+      "loss": 0.0146,
+      "step": 13889
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012130348907148994,
+      "loss": 0.0041,
+      "step": 13890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001212844619982077,
+      "loss": 0.0117,
+      "step": 13891
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012126543540452924,
+      "loss": 0.0084,
+      "step": 13892
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012124640929077222,
+      "loss": 0.0129,
+      "step": 13893
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012122738365725439,
+      "loss": 0.0097,
+      "step": 13894
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012120835850429363,
+      "loss": 0.0156,
+      "step": 13895
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012118933383220757,
+      "loss": 0.0154,
+      "step": 13896
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012117030964131409,
+      "loss": 0.0048,
+      "step": 13897
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012115128593193081,
+      "loss": 0.0042,
+      "step": 13898
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012113226270437551,
+      "loss": 0.0023,
+      "step": 13899
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012111323995896592,
+      "loss": 0.0038,
+      "step": 13900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012109421769601972,
+      "loss": 0.0023,
+      "step": 13901
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012107519591585472,
+      "loss": 0.016,
+      "step": 13902
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012105617461878853,
+      "loss": 0.0052,
+      "step": 13903
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001210371538051388,
+      "loss": 0.0134,
+      "step": 13904
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012101813347522335,
+      "loss": 0.0223,
+      "step": 13905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001209991136293597,
+      "loss": 0.0169,
+      "step": 13906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012098009426786567,
+      "loss": 0.0082,
+      "step": 13907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001209610753910588,
+      "loss": 0.0257,
+      "step": 13908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012094205699925679,
+      "loss": 0.033,
+      "step": 13909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001209230390927773,
+      "loss": 0.0008,
+      "step": 13910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012090402167193795,
+      "loss": 0.0004,
+      "step": 13911
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012088500473705629,
+      "loss": 0.0072,
+      "step": 13912
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012086598828845009,
+      "loss": 0.0065,
+      "step": 13913
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012084697232643683,
+      "loss": 0.0149,
+      "step": 13914
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001208279568513342,
+      "loss": 0.0152,
+      "step": 13915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012080894186345977,
+      "loss": 0.0028,
+      "step": 13916
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012078992736313104,
+      "loss": 0.0092,
+      "step": 13917
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012077091335066574,
+      "loss": 0.0051,
+      "step": 13918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012075189982638128,
+      "loss": 0.0143,
+      "step": 13919
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012073288679059535,
+      "loss": 0.0074,
+      "step": 13920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012071387424362546,
+      "loss": 0.0074,
+      "step": 13921
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012069486218578912,
+      "loss": 0.0059,
+      "step": 13922
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012067585061740395,
+      "loss": 0.0186,
+      "step": 13923
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012065683953878737,
+      "loss": 0.0105,
+      "step": 13924
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000120637828950257,
+      "loss": 0.0039,
+      "step": 13925
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012061881885213035,
+      "loss": 0.0093,
+      "step": 13926
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012059980924472478,
+      "loss": 0.0021,
+      "step": 13927
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000120580800128358,
+      "loss": 0.0025,
+      "step": 13928
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001205617915033473,
+      "loss": 0.0038,
+      "step": 13929
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012054278337001027,
+      "loss": 0.0099,
+      "step": 13930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012052377572866443,
+      "loss": 0.0092,
+      "step": 13931
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012050476857962708,
+      "loss": 0.0048,
+      "step": 13932
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012048576192321584,
+      "loss": 0.0071,
+      "step": 13933
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001204667557597481,
+      "loss": 0.0087,
+      "step": 13934
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012044775008954123,
+      "loss": 0.0145,
+      "step": 13935
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012042874491291276,
+      "loss": 0.0392,
+      "step": 13936
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012040974023018001,
+      "loss": 0.0093,
+      "step": 13937
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012039073604166054,
+      "loss": 0.009,
+      "step": 13938
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001203717323476716,
+      "loss": 0.005,
+      "step": 13939
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012035272914853069,
+      "loss": 0.0127,
+      "step": 13940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012033372644455516,
+      "loss": 0.009,
+      "step": 13941
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012031472423606234,
+      "loss": 0.0119,
+      "step": 13942
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012029572252336976,
+      "loss": 0.0309,
+      "step": 13943
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012027672130679466,
+      "loss": 0.0077,
+      "step": 13944
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012025772058665436,
+      "loss": 0.006,
+      "step": 13945
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012023872036326635,
+      "loss": 0.0215,
+      "step": 13946
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012021972063694782,
+      "loss": 0.0049,
+      "step": 13947
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012020072140801622,
+      "loss": 0.0379,
+      "step": 13948
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012018172267678879,
+      "loss": 0.0089,
+      "step": 13949
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012016272444358289,
+      "loss": 0.0072,
+      "step": 13950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001201437267087158,
+      "loss": 0.0091,
+      "step": 13951
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012012472947250484,
+      "loss": 0.0014,
+      "step": 13952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012010573273526728,
+      "loss": 0.0029,
+      "step": 13953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012008673649732046,
+      "loss": 0.0053,
+      "step": 13954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012006774075898154,
+      "loss": 0.016,
+      "step": 13955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001200487455205679,
+      "loss": 0.0006,
+      "step": 13956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001200297507823967,
+      "loss": 0.0098,
+      "step": 13957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001200107565447853,
+      "loss": 0.0048,
+      "step": 13958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011999176280805087,
+      "loss": 0.0146,
+      "step": 13959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011997276957251056,
+      "loss": 0.0069,
+      "step": 13960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011995377683848173,
+      "loss": 0.0113,
+      "step": 13961
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011993478460628158,
+      "loss": 0.0017,
+      "step": 13962
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011991579287622722,
+      "loss": 0.0088,
+      "step": 13963
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011989680164863596,
+      "loss": 0.0069,
+      "step": 13964
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011987781092382487,
+      "loss": 0.005,
+      "step": 13965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011985882070211127,
+      "loss": 0.0056,
+      "step": 13966
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011983983098381224,
+      "loss": 0.0061,
+      "step": 13967
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011982084176924493,
+      "loss": 0.0065,
+      "step": 13968
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011980185305872656,
+      "loss": 0.0104,
+      "step": 13969
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011978286485257421,
+      "loss": 0.0088,
+      "step": 13970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011976387715110511,
+      "loss": 0.0062,
+      "step": 13971
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011974488995463631,
+      "loss": 0.0109,
+      "step": 13972
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001197259032634849,
+      "loss": 0.0057,
+      "step": 13973
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011970691707796815,
+      "loss": 0.012,
+      "step": 13974
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011968793139840298,
+      "loss": 0.0081,
+      "step": 13975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011966894622510666,
+      "loss": 0.0039,
+      "step": 13976
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011964996155839617,
+      "loss": 0.0059,
+      "step": 13977
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011963097739858857,
+      "loss": 0.0078,
+      "step": 13978
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011961199374600102,
+      "loss": 0.0126,
+      "step": 13979
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011959301060095047,
+      "loss": 0.0012,
+      "step": 13980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011957402796375412,
+      "loss": 0.0038,
+      "step": 13981
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011955504583472889,
+      "loss": 0.003,
+      "step": 13982
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011953606421419184,
+      "loss": 0.0085,
+      "step": 13983
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011951708310246008,
+      "loss": 0.0216,
+      "step": 13984
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011949810249985057,
+      "loss": 0.0052,
+      "step": 13985
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011947912240668028,
+      "loss": 0.0119,
+      "step": 13986
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011946014282326629,
+      "loss": 0.02,
+      "step": 13987
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011944116374992552,
+      "loss": 0.0057,
+      "step": 13988
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011942218518697506,
+      "loss": 0.0008,
+      "step": 13989
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011940320713473178,
+      "loss": 0.0012,
+      "step": 13990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011938422959351272,
+      "loss": 0.0084,
+      "step": 13991
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011936525256363481,
+      "loss": 0.0031,
+      "step": 13992
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011934627604541501,
+      "loss": 0.0161,
+      "step": 13993
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011932730003917026,
+      "loss": 0.0088,
+      "step": 13994
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011930832454521753,
+      "loss": 0.0142,
+      "step": 13995
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011928934956387365,
+      "loss": 0.0032,
+      "step": 13996
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011927037509545567,
+      "loss": 0.0183,
+      "step": 13997
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011925140114028038,
+      "loss": 0.0046,
+      "step": 13998
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011923242769866477,
+      "loss": 0.0184,
+      "step": 13999
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011921345477092572,
+      "loss": 0.007,
+      "step": 14000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011919448235738,
+      "loss": 0.0019,
+      "step": 14001
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011917551045834463,
+      "loss": 0.016,
+      "step": 14002
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011915653907413641,
+      "loss": 0.0114,
+      "step": 14003
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001191375682050722,
+      "loss": 0.0118,
+      "step": 14004
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011911859785146891,
+      "loss": 0.0119,
+      "step": 14005
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011909962801364324,
+      "loss": 0.0073,
+      "step": 14006
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011908065869191219,
+      "loss": 0.0014,
+      "step": 14007
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001190616898865925,
+      "loss": 0.0073,
+      "step": 14008
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011904272159800093,
+      "loss": 0.0053,
+      "step": 14009
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011902375382645441,
+      "loss": 0.0084,
+      "step": 14010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011900478657226961,
+      "loss": 0.0038,
+      "step": 14011
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011898581983576345,
+      "loss": 0.0158,
+      "step": 14012
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001189668536172526,
+      "loss": 0.0017,
+      "step": 14013
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011894788791705385,
+      "loss": 0.0101,
+      "step": 14014
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011892892273548403,
+      "loss": 0.0273,
+      "step": 14015
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011890995807285981,
+      "loss": 0.0065,
+      "step": 14016
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011889099392949804,
+      "loss": 0.0024,
+      "step": 14017
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011887203030571539,
+      "loss": 0.0073,
+      "step": 14018
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011885306720182852,
+      "loss": 0.0082,
+      "step": 14019
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011883410461815429,
+      "loss": 0.0052,
+      "step": 14020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011881514255500928,
+      "loss": 0.0058,
+      "step": 14021
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011879618101271032,
+      "loss": 0.0134,
+      "step": 14022
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011877721999157401,
+      "loss": 0.0095,
+      "step": 14023
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011875825949191704,
+      "loss": 0.0064,
+      "step": 14024
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011873929951405614,
+      "loss": 0.0054,
+      "step": 14025
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001187203400583079,
+      "loss": 0.0025,
+      "step": 14026
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011870138112498909,
+      "loss": 0.0034,
+      "step": 14027
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011868242271441629,
+      "loss": 0.0085,
+      "step": 14028
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011866346482690609,
+      "loss": 0.0115,
+      "step": 14029
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011864450746277526,
+      "loss": 0.0036,
+      "step": 14030
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011862555062234025,
+      "loss": 0.0098,
+      "step": 14031
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011860659430591786,
+      "loss": 0.0068,
+      "step": 14032
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011858763851382454,
+      "loss": 0.0122,
+      "step": 14033
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011856868324637698,
+      "loss": 0.0108,
+      "step": 14034
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011854972850389173,
+      "loss": 0.0218,
+      "step": 14035
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001185307742866854,
+      "loss": 0.0164,
+      "step": 14036
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001185118205950745,
+      "loss": 0.0112,
+      "step": 14037
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011849286742937569,
+      "loss": 0.0058,
+      "step": 14038
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001184739147899054,
+      "loss": 0.0126,
+      "step": 14039
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011845496267698032,
+      "loss": 0.0129,
+      "step": 14040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011843601109091688,
+      "loss": 0.003,
+      "step": 14041
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001184170600320316,
+      "loss": 0.0076,
+      "step": 14042
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011839810950064109,
+      "loss": 0.0214,
+      "step": 14043
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011837915949706173,
+      "loss": 0.0111,
+      "step": 14044
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011836021002161013,
+      "loss": 0.0089,
+      "step": 14045
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011834126107460278,
+      "loss": 0.0025,
+      "step": 14046
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011832231265635607,
+      "loss": 0.0097,
+      "step": 14047
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011830336476718661,
+      "loss": 0.0104,
+      "step": 14048
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011828441740741071,
+      "loss": 0.0233,
+      "step": 14049
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011826547057734497,
+      "loss": 0.014,
+      "step": 14050
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011824652427730578,
+      "loss": 0.0173,
+      "step": 14051
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011822757850760952,
+      "loss": 0.0221,
+      "step": 14052
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011820863326857273,
+      "loss": 0.002,
+      "step": 14053
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001181896885605117,
+      "loss": 0.0177,
+      "step": 14054
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011817074438374296,
+      "loss": 0.0059,
+      "step": 14055
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001181518007385829,
+      "loss": 0.0037,
+      "step": 14056
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011813285762534783,
+      "loss": 0.0142,
+      "step": 14057
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011811391504435422,
+      "loss": 0.0089,
+      "step": 14058
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011809497299591847,
+      "loss": 0.0093,
+      "step": 14059
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001180760314803568,
+      "loss": 0.0117,
+      "step": 14060
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011805709049798573,
+      "loss": 0.0017,
+      "step": 14061
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001180381500491215,
+      "loss": 0.0036,
+      "step": 14062
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011801921013408054,
+      "loss": 0.0058,
+      "step": 14063
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011800027075317911,
+      "loss": 0.0196,
+      "step": 14064
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011798133190673355,
+      "loss": 0.0038,
+      "step": 14065
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011796239359506021,
+      "loss": 0.0016,
+      "step": 14066
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011794345581847534,
+      "loss": 0.0072,
+      "step": 14067
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011792451857729533,
+      "loss": 0.0016,
+      "step": 14068
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001179055818718364,
+      "loss": 0.0022,
+      "step": 14069
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011788664570241477,
+      "loss": 0.0062,
+      "step": 14070
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011786771006934686,
+      "loss": 0.0109,
+      "step": 14071
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011784877497294876,
+      "loss": 0.0133,
+      "step": 14072
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011782984041353686,
+      "loss": 0.0093,
+      "step": 14073
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011781090639142734,
+      "loss": 0.005,
+      "step": 14074
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011779197290693645,
+      "loss": 0.0143,
+      "step": 14075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011777303996038038,
+      "loss": 0.0111,
+      "step": 14076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001177541075520754,
+      "loss": 0.0213,
+      "step": 14077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011773517568233769,
+      "loss": 0.0102,
+      "step": 14078
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011771624435148347,
+      "loss": 0.006,
+      "step": 14079
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011769731355982887,
+      "loss": 0.0143,
+      "step": 14080
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011767838330769015,
+      "loss": 0.0036,
+      "step": 14081
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011765945359538339,
+      "loss": 0.0077,
+      "step": 14082
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011764052442322485,
+      "loss": 0.0045,
+      "step": 14083
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011762159579153064,
+      "loss": 0.0223,
+      "step": 14084
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011760266770061683,
+      "loss": 0.0083,
+      "step": 14085
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011758374015079969,
+      "loss": 0.0104,
+      "step": 14086
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011756481314239527,
+      "loss": 0.0086,
+      "step": 14087
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011754588667571965,
+      "loss": 0.009,
+      "step": 14088
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011752696075108905,
+      "loss": 0.006,
+      "step": 14089
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011750803536881943,
+      "loss": 0.004,
+      "step": 14090
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011748911052922702,
+      "loss": 0.0047,
+      "step": 14091
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011747018623262784,
+      "loss": 0.0089,
+      "step": 14092
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011745126247933788,
+      "loss": 0.0022,
+      "step": 14093
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011743233926967335,
+      "loss": 0.024,
+      "step": 14094
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011741341660395016,
+      "loss": 0.0044,
+      "step": 14095
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011739449448248448,
+      "loss": 0.0012,
+      "step": 14096
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011737557290559229,
+      "loss": 0.0037,
+      "step": 14097
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011735665187358955,
+      "loss": 0.0237,
+      "step": 14098
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011733773138679242,
+      "loss": 0.0175,
+      "step": 14099
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011731881144551675,
+      "loss": 0.0131,
+      "step": 14100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001172998920500787,
+      "loss": 0.0042,
+      "step": 14101
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011728097320079418,
+      "loss": 0.0074,
+      "step": 14102
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011726205489797911,
+      "loss": 0.0053,
+      "step": 14103
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011724313714194957,
+      "loss": 0.0034,
+      "step": 14104
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011722421993302141,
+      "loss": 0.0039,
+      "step": 14105
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011720530327151071,
+      "loss": 0.0041,
+      "step": 14106
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011718638715773332,
+      "loss": 0.0072,
+      "step": 14107
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011716747159200516,
+      "loss": 0.0075,
+      "step": 14108
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011714855657464229,
+      "loss": 0.0172,
+      "step": 14109
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011712964210596052,
+      "loss": 0.0036,
+      "step": 14110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001171107281862757,
+      "loss": 0.008,
+      "step": 14111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011709181481590388,
+      "loss": 0.0041,
+      "step": 14112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011707290199516083,
+      "loss": 0.0064,
+      "step": 14113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001170539897243625,
+      "loss": 0.0028,
+      "step": 14114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011703507800382472,
+      "loss": 0.0045,
+      "step": 14115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011701616683386335,
+      "loss": 0.0069,
+      "step": 14116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011699725621479426,
+      "loss": 0.006,
+      "step": 14117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001169783461469333,
+      "loss": 0.0144,
+      "step": 14118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011695943663059627,
+      "loss": 0.0071,
+      "step": 14119
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011694052766609906,
+      "loss": 0.005,
+      "step": 14120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011692161925375738,
+      "loss": 0.0101,
+      "step": 14121
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011690271139388714,
+      "loss": 0.0048,
+      "step": 14122
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011688380408680407,
+      "loss": 0.0161,
+      "step": 14123
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011686489733282404,
+      "loss": 0.008,
+      "step": 14124
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011684599113226274,
+      "loss": 0.0062,
+      "step": 14125
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011682708548543592,
+      "loss": 0.0074,
+      "step": 14126
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011680818039265944,
+      "loss": 0.0072,
+      "step": 14127
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011678927585424899,
+      "loss": 0.0132,
+      "step": 14128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011677037187052033,
+      "loss": 0.0319,
+      "step": 14129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001167514684417892,
+      "loss": 0.0171,
+      "step": 14130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011673256556837125,
+      "loss": 0.0222,
+      "step": 14131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011671366325058234,
+      "loss": 0.0178,
+      "step": 14132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011669476148873805,
+      "loss": 0.006,
+      "step": 14133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011667586028315407,
+      "loss": 0.0049,
+      "step": 14134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011665695963414618,
+      "loss": 0.0079,
+      "step": 14135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011663805954202995,
+      "loss": 0.0284,
+      "step": 14136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011661916000712115,
+      "loss": 0.0101,
+      "step": 14137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011660026102973536,
+      "loss": 0.008,
+      "step": 14138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011658136261018824,
+      "loss": 0.0037,
+      "step": 14139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001165624647487955,
+      "loss": 0.0061,
+      "step": 14140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011654356744587265,
+      "loss": 0.0057,
+      "step": 14141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011652467070173545,
+      "loss": 0.0131,
+      "step": 14142
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011650577451669944,
+      "loss": 0.0041,
+      "step": 14143
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011648687889108014,
+      "loss": 0.0025,
+      "step": 14144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011646798382519332,
+      "loss": 0.0044,
+      "step": 14145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011644908931935436,
+      "loss": 0.004,
+      "step": 14146
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011643019537387903,
+      "loss": 0.009,
+      "step": 14147
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011641130198908277,
+      "loss": 0.007,
+      "step": 14148
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011639240916528116,
+      "loss": 0.0099,
+      "step": 14149
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011637351690278976,
+      "loss": 0.0021,
+      "step": 14150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011635462520192408,
+      "loss": 0.0167,
+      "step": 14151
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011633573406299972,
+      "loss": 0.0021,
+      "step": 14152
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011631684348633216,
+      "loss": 0.0169,
+      "step": 14153
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011629795347223682,
+      "loss": 0.0015,
+      "step": 14154
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011627906402102934,
+      "loss": 0.0117,
+      "step": 14155
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011626017513302507,
+      "loss": 0.0137,
+      "step": 14156
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011624128680853963,
+      "loss": 0.0093,
+      "step": 14157
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011622239904788843,
+      "loss": 0.0032,
+      "step": 14158
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011620351185138688,
+      "loss": 0.0094,
+      "step": 14159
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001161846252193505,
+      "loss": 0.0116,
+      "step": 14160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011616573915209474,
+      "loss": 0.0041,
+      "step": 14161
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011614685364993493,
+      "loss": 0.0013,
+      "step": 14162
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011612796871318663,
+      "loss": 0.0153,
+      "step": 14163
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011610908434216514,
+      "loss": 0.0089,
+      "step": 14164
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011609020053718597,
+      "loss": 0.0085,
+      "step": 14165
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011607131729856446,
+      "loss": 0.0024,
+      "step": 14166
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011605243462661594,
+      "loss": 0.0021,
+      "step": 14167
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011603355252165586,
+      "loss": 0.0187,
+      "step": 14168
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011601467098399958,
+      "loss": 0.0058,
+      "step": 14169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011599579001396245,
+      "loss": 0.0215,
+      "step": 14170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011597690961185984,
+      "loss": 0.0064,
+      "step": 14171
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000115958029778007,
+      "loss": 0.0022,
+      "step": 14172
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011593915051271937,
+      "loss": 0.0087,
+      "step": 14173
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011592027181631218,
+      "loss": 0.0011,
+      "step": 14174
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011590139368910084,
+      "loss": 0.0029,
+      "step": 14175
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011588251613140058,
+      "loss": 0.0164,
+      "step": 14176
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011586363914352665,
+      "loss": 0.0046,
+      "step": 14177
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011584476272579444,
+      "loss": 0.0064,
+      "step": 14178
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001158258868785191,
+      "loss": 0.0015,
+      "step": 14179
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011580701160201599,
+      "loss": 0.0097,
+      "step": 14180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011578813689660038,
+      "loss": 0.0065,
+      "step": 14181
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011576926276258736,
+      "loss": 0.0095,
+      "step": 14182
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011575038920029236,
+      "loss": 0.0175,
+      "step": 14183
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001157315162100305,
+      "loss": 0.0049,
+      "step": 14184
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011571264379211695,
+      "loss": 0.0115,
+      "step": 14185
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011569377194686704,
+      "loss": 0.0004,
+      "step": 14186
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011567490067459582,
+      "loss": 0.0025,
+      "step": 14187
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011565602997561862,
+      "loss": 0.0071,
+      "step": 14188
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011563715985025051,
+      "loss": 0.0098,
+      "step": 14189
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011561829029880671,
+      "loss": 0.0033,
+      "step": 14190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011559942132160235,
+      "loss": 0.002,
+      "step": 14191
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011558055291895255,
+      "loss": 0.01,
+      "step": 14192
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011556168509117257,
+      "loss": 0.0038,
+      "step": 14193
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011554281783857745,
+      "loss": 0.0075,
+      "step": 14194
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011552395116148226,
+      "loss": 0.0059,
+      "step": 14195
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011550508506020221,
+      "loss": 0.0092,
+      "step": 14196
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011548621953505229,
+      "loss": 0.0009,
+      "step": 14197
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011546735458634772,
+      "loss": 0.0132,
+      "step": 14198
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011544849021440349,
+      "loss": 0.0038,
+      "step": 14199
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011542962641953469,
+      "loss": 0.0065,
+      "step": 14200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011541076320205637,
+      "loss": 0.0083,
+      "step": 14201
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011539190056228362,
+      "loss": 0.0147,
+      "step": 14202
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011537303850053144,
+      "loss": 0.0114,
+      "step": 14203
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011535417701711494,
+      "loss": 0.0026,
+      "step": 14204
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011533531611234898,
+      "loss": 0.0024,
+      "step": 14205
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011531645578654875,
+      "loss": 0.0047,
+      "step": 14206
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001152975960400292,
+      "loss": 0.0022,
+      "step": 14207
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011527873687310522,
+      "loss": 0.0098,
+      "step": 14208
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011525987828609193,
+      "loss": 0.0023,
+      "step": 14209
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011524102027930421,
+      "loss": 0.0006,
+      "step": 14210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011522216285305708,
+      "loss": 0.0027,
+      "step": 14211
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011520330600766552,
+      "loss": 0.009,
+      "step": 14212
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011518444974344435,
+      "loss": 0.0061,
+      "step": 14213
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011516559406070866,
+      "loss": 0.0092,
+      "step": 14214
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011514673895977327,
+      "loss": 0.0234,
+      "step": 14215
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011512788444095316,
+      "loss": 0.0031,
+      "step": 14216
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001151090305045632,
+      "loss": 0.0037,
+      "step": 14217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011509017715091824,
+      "loss": 0.0066,
+      "step": 14218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011507132438033329,
+      "loss": 0.0156,
+      "step": 14219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011505247219312309,
+      "loss": 0.0013,
+      "step": 14220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011503362058960261,
+      "loss": 0.0085,
+      "step": 14221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011501476957008667,
+      "loss": 0.0147,
+      "step": 14222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011499591913489008,
+      "loss": 0.0035,
+      "step": 14223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011497706928432777,
+      "loss": 0.0107,
+      "step": 14224
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011495822001871448,
+      "loss": 0.0064,
+      "step": 14225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011493937133836509,
+      "loss": 0.0212,
+      "step": 14226
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011492052324359437,
+      "loss": 0.003,
+      "step": 14227
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001149016757347171,
+      "loss": 0.0141,
+      "step": 14228
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011488282881204816,
+      "loss": 0.0072,
+      "step": 14229
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011486398247590219,
+      "loss": 0.013,
+      "step": 14230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011484513672659409,
+      "loss": 0.0029,
+      "step": 14231
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011482629156443856,
+      "loss": 0.0051,
+      "step": 14232
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001148074469897503,
+      "loss": 0.0204,
+      "step": 14233
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011478860300284418,
+      "loss": 0.0076,
+      "step": 14234
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011476975960403484,
+      "loss": 0.0101,
+      "step": 14235
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011475091679363698,
+      "loss": 0.0088,
+      "step": 14236
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001147320745719654,
+      "loss": 0.0062,
+      "step": 14237
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011471323293933468,
+      "loss": 0.01,
+      "step": 14238
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011469439189605964,
+      "loss": 0.0123,
+      "step": 14239
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011467555144245486,
+      "loss": 0.0061,
+      "step": 14240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011465671157883508,
+      "loss": 0.0201,
+      "step": 14241
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011463787230551491,
+      "loss": 0.0066,
+      "step": 14242
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011461903362280903,
+      "loss": 0.004,
+      "step": 14243
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011460019553103206,
+      "loss": 0.0091,
+      "step": 14244
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001145813580304987,
+      "loss": 0.0033,
+      "step": 14245
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011456252112152345,
+      "loss": 0.029,
+      "step": 14246
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011454368480442103,
+      "loss": 0.0043,
+      "step": 14247
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011452484907950597,
+      "loss": 0.0139,
+      "step": 14248
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011450601394709294,
+      "loss": 0.011,
+      "step": 14249
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001144871794074965,
+      "loss": 0.0205,
+      "step": 14250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011446834546103111,
+      "loss": 0.0134,
+      "step": 14251
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011444951210801146,
+      "loss": 0.0079,
+      "step": 14252
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011443067934875206,
+      "loss": 0.0044,
+      "step": 14253
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011441184718356747,
+      "loss": 0.0048,
+      "step": 14254
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011439301561277222,
+      "loss": 0.0051,
+      "step": 14255
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011437418463668078,
+      "loss": 0.0057,
+      "step": 14256
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011435535425560777,
+      "loss": 0.0083,
+      "step": 14257
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011433652446986761,
+      "loss": 0.0108,
+      "step": 14258
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011431769527977476,
+      "loss": 0.0078,
+      "step": 14259
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011429886668564382,
+      "loss": 0.0033,
+      "step": 14260
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011428003868778912,
+      "loss": 0.0066,
+      "step": 14261
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011426121128652526,
+      "loss": 0.0108,
+      "step": 14262
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011424238448216661,
+      "loss": 0.0146,
+      "step": 14263
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011422355827502761,
+      "loss": 0.0076,
+      "step": 14264
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011420473266542279,
+      "loss": 0.0069,
+      "step": 14265
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011418590765366642,
+      "loss": 0.0065,
+      "step": 14266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011416708324007307,
+      "loss": 0.0045,
+      "step": 14267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011414825942495708,
+      "loss": 0.0031,
+      "step": 14268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011412943620863275,
+      "loss": 0.0219,
+      "step": 14269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011411061359141462,
+      "loss": 0.0002,
+      "step": 14270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001140917915736169,
+      "loss": 0.0054,
+      "step": 14271
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011407297015555416,
+      "loss": 0.0076,
+      "step": 14272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011405414933754056,
+      "loss": 0.0064,
+      "step": 14273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001140353291198905,
+      "loss": 0.0007,
+      "step": 14274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011401650950291839,
+      "loss": 0.0019,
+      "step": 14275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011399769048693843,
+      "loss": 0.0117,
+      "step": 14276
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011397887207226508,
+      "loss": 0.0161,
+      "step": 14277
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011396005425921253,
+      "loss": 0.01,
+      "step": 14278
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011394123704809507,
+      "loss": 0.0074,
+      "step": 14279
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011392242043922706,
+      "loss": 0.0036,
+      "step": 14280
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011390360443292275,
+      "loss": 0.0013,
+      "step": 14281
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001138847890294963,
+      "loss": 0.0026,
+      "step": 14282
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011386597422926209,
+      "loss": 0.015,
+      "step": 14283
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011384716003253431,
+      "loss": 0.006,
+      "step": 14284
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011382834643962721,
+      "loss": 0.0085,
+      "step": 14285
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011380953345085503,
+      "loss": 0.0063,
+      "step": 14286
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001137907210665319,
+      "loss": 0.0066,
+      "step": 14287
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011377190928697213,
+      "loss": 0.0092,
+      "step": 14288
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011375309811248979,
+      "loss": 0.0059,
+      "step": 14289
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001137342875433992,
+      "loss": 0.0071,
+      "step": 14290
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011371547758001448,
+      "loss": 0.0275,
+      "step": 14291
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011369666822264972,
+      "loss": 0.0214,
+      "step": 14292
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011367785947161912,
+      "loss": 0.0095,
+      "step": 14293
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011365905132723684,
+      "loss": 0.013,
+      "step": 14294
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011364024378981703,
+      "loss": 0.0092,
+      "step": 14295
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001136214368596738,
+      "loss": 0.0022,
+      "step": 14296
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011360263053712119,
+      "loss": 0.0209,
+      "step": 14297
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001135838248224734,
+      "loss": 0.0054,
+      "step": 14298
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011356501971604442,
+      "loss": 0.0117,
+      "step": 14299
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011354621521814845,
+      "loss": 0.003,
+      "step": 14300
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011352741132909948,
+      "loss": 0.0057,
+      "step": 14301
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011350860804921155,
+      "loss": 0.0054,
+      "step": 14302
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001134898053787988,
+      "loss": 0.0053,
+      "step": 14303
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011347100331817516,
+      "loss": 0.011,
+      "step": 14304
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011345220186765475,
+      "loss": 0.0111,
+      "step": 14305
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011343340102755158,
+      "loss": 0.0174,
+      "step": 14306
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001134146007981796,
+      "loss": 0.0029,
+      "step": 14307
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011339580117985287,
+      "loss": 0.0037,
+      "step": 14308
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011337700217288538,
+      "loss": 0.0025,
+      "step": 14309
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011335820377759101,
+      "loss": 0.0053,
+      "step": 14310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011333940599428388,
+      "loss": 0.0011,
+      "step": 14311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001133206088232778,
+      "loss": 0.0107,
+      "step": 14312
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011330181226488685,
+      "loss": 0.0049,
+      "step": 14313
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011328301631942489,
+      "loss": 0.0141,
+      "step": 14314
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011326422098720585,
+      "loss": 0.0025,
+      "step": 14315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011324542626854367,
+      "loss": 0.0048,
+      "step": 14316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011322663216375223,
+      "loss": 0.0079,
+      "step": 14317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011320783867314549,
+      "loss": 0.0023,
+      "step": 14318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011318904579703731,
+      "loss": 0.0122,
+      "step": 14319
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011317025353574149,
+      "loss": 0.0238,
+      "step": 14320
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000113151461889572,
+      "loss": 0.005,
+      "step": 14321
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001131326708588426,
+      "loss": 0.0028,
+      "step": 14322
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011311388044386725,
+      "loss": 0.0127,
+      "step": 14323
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011309509064495969,
+      "loss": 0.0119,
+      "step": 14324
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011307630146243378,
+      "loss": 0.0121,
+      "step": 14325
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011305751289660335,
+      "loss": 0.0132,
+      "step": 14326
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011303872494778214,
+      "loss": 0.0036,
+      "step": 14327
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011301993761628408,
+      "loss": 0.0042,
+      "step": 14328
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011300115090242284,
+      "loss": 0.0061,
+      "step": 14329
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011298236480651217,
+      "loss": 0.0106,
+      "step": 14330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011296357932886594,
+      "loss": 0.0017,
+      "step": 14331
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011294479446979785,
+      "loss": 0.0002,
+      "step": 14332
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011292601022962156,
+      "loss": 0.0142,
+      "step": 14333
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011290722660865097,
+      "loss": 0.0206,
+      "step": 14334
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011288844360719965,
+      "loss": 0.0024,
+      "step": 14335
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011286966122558138,
+      "loss": 0.0236,
+      "step": 14336
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011285087946410989,
+      "loss": 0.0169,
+      "step": 14337
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001128320983230988,
+      "loss": 0.0221,
+      "step": 14338
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011281331780286187,
+      "loss": 0.0048,
+      "step": 14339
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011279453790371268,
+      "loss": 0.0125,
+      "step": 14340
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011277575862596499,
+      "loss": 0.0063,
+      "step": 14341
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011275697996993241,
+      "loss": 0.0113,
+      "step": 14342
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011273820193592849,
+      "loss": 0.0038,
+      "step": 14343
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011271942452426698,
+      "loss": 0.0024,
+      "step": 14344
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011270064773526141,
+      "loss": 0.019,
+      "step": 14345
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011268187156922547,
+      "loss": 0.0019,
+      "step": 14346
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011266309602647273,
+      "loss": 0.0082,
+      "step": 14347
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011264432110731671,
+      "loss": 0.0299,
+      "step": 14348
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001126255468120711,
+      "loss": 0.0089,
+      "step": 14349
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011260677314104935,
+      "loss": 0.0147,
+      "step": 14350
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011258800009456514,
+      "loss": 0.0094,
+      "step": 14351
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011256922767293195,
+      "loss": 0.0033,
+      "step": 14352
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011255045587646323,
+      "loss": 0.0075,
+      "step": 14353
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011253168470547268,
+      "loss": 0.0083,
+      "step": 14354
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011251291416027367,
+      "loss": 0.0064,
+      "step": 14355
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011249414424117975,
+      "loss": 0.0022,
+      "step": 14356
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011247537494850444,
+      "loss": 0.0004,
+      "step": 14357
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011245660628256115,
+      "loss": 0.0047,
+      "step": 14358
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001124378382436635,
+      "loss": 0.0103,
+      "step": 14359
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011241907083212483,
+      "loss": 0.0094,
+      "step": 14360
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011240030404825854,
+      "loss": 0.0102,
+      "step": 14361
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011238153789237821,
+      "loss": 0.0075,
+      "step": 14362
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011236277236479717,
+      "loss": 0.0022,
+      "step": 14363
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011234400746582892,
+      "loss": 0.0028,
+      "step": 14364
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011232524319578681,
+      "loss": 0.0149,
+      "step": 14365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011230647955498424,
+      "loss": 0.0016,
+      "step": 14366
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011228771654373464,
+      "loss": 0.0063,
+      "step": 14367
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011226895416235133,
+      "loss": 0.0041,
+      "step": 14368
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011225019241114772,
+      "loss": 0.0087,
+      "step": 14369
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011223143129043719,
+      "loss": 0.0049,
+      "step": 14370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011221267080053297,
+      "loss": 0.0139,
+      "step": 14371
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011219391094174856,
+      "loss": 0.0042,
+      "step": 14372
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011217515171439714,
+      "loss": 0.0055,
+      "step": 14373
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011215639311879212,
+      "loss": 0.0186,
+      "step": 14374
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011213763515524681,
+      "loss": 0.0071,
+      "step": 14375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011211887782407436,
+      "loss": 0.0075,
+      "step": 14376
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011210012112558821,
+      "loss": 0.0226,
+      "step": 14377
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011208136506010157,
+      "loss": 0.0056,
+      "step": 14378
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011206260962792773,
+      "loss": 0.0062,
+      "step": 14379
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011204385482937993,
+      "loss": 0.0219,
+      "step": 14380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011202510066477134,
+      "loss": 0.0081,
+      "step": 14381
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011200634713441533,
+      "loss": 0.0076,
+      "step": 14382
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011198759423862503,
+      "loss": 0.0067,
+      "step": 14383
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011196884197771358,
+      "loss": 0.0166,
+      "step": 14384
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011195009035199436,
+      "loss": 0.0104,
+      "step": 14385
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011193133936178037,
+      "loss": 0.0145,
+      "step": 14386
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011191258900738493,
+      "loss": 0.0109,
+      "step": 14387
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001118938392891211,
+      "loss": 0.001,
+      "step": 14388
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011187509020730209,
+      "loss": 0.0316,
+      "step": 14389
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011185634176224111,
+      "loss": 0.0046,
+      "step": 14390
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011183759395425113,
+      "loss": 0.0112,
+      "step": 14391
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011181884678364543,
+      "loss": 0.0029,
+      "step": 14392
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011180010025073708,
+      "loss": 0.007,
+      "step": 14393
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011178135435583908,
+      "loss": 0.0101,
+      "step": 14394
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011176260909926468,
+      "loss": 0.0135,
+      "step": 14395
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011174386448132683,
+      "loss": 0.0213,
+      "step": 14396
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011172512050233872,
+      "loss": 0.0073,
+      "step": 14397
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011170637716261329,
+      "loss": 0.0055,
+      "step": 14398
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011168763446246365,
+      "loss": 0.0028,
+      "step": 14399
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011166889240220286,
+      "loss": 0.0061,
+      "step": 14400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011165015098214389,
+      "loss": 0.0098,
+      "step": 14401
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011163141020259985,
+      "loss": 0.0053,
+      "step": 14402
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011161267006388367,
+      "loss": 0.0051,
+      "step": 14403
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011159393056630828,
+      "loss": 0.0031,
+      "step": 14404
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011157519171018686,
+      "loss": 0.0075,
+      "step": 14405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011155645349583221,
+      "loss": 0.0058,
+      "step": 14406
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011153771592355735,
+      "loss": 0.0168,
+      "step": 14407
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011151897899367524,
+      "loss": 0.0058,
+      "step": 14408
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011150024270649883,
+      "loss": 0.01,
+      "step": 14409
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011148150706234102,
+      "loss": 0.0071,
+      "step": 14410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001114627720615148,
+      "loss": 0.0063,
+      "step": 14411
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011144403770433295,
+      "loss": 0.0108,
+      "step": 14412
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011142530399110853,
+      "loss": 0.0217,
+      "step": 14413
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011140657092215429,
+      "loss": 0.0067,
+      "step": 14414
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011138783849778322,
+      "loss": 0.0075,
+      "step": 14415
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011136910671830814,
+      "loss": 0.004,
+      "step": 14416
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011135037558404182,
+      "loss": 0.012,
+      "step": 14417
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001113316450952972,
+      "loss": 0.0072,
+      "step": 14418
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011131291525238715,
+      "loss": 0.0038,
+      "step": 14419
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001112941860556244,
+      "loss": 0.0022,
+      "step": 14420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011127545750532186,
+      "loss": 0.0202,
+      "step": 14421
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011125672960179218,
+      "loss": 0.0062,
+      "step": 14422
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011123800234534834,
+      "loss": 0.0029,
+      "step": 14423
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011121927573630298,
+      "loss": 0.0045,
+      "step": 14424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011120054977496895,
+      "loss": 0.0036,
+      "step": 14425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000111181824461659,
+      "loss": 0.0047,
+      "step": 14426
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011116309979668579,
+      "loss": 0.0035,
+      "step": 14427
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011114437578036218,
+      "loss": 0.016,
+      "step": 14428
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001111256524130008,
+      "loss": 0.0042,
+      "step": 14429
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011110692969491439,
+      "loss": 0.0074,
+      "step": 14430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011108820762641573,
+      "loss": 0.002,
+      "step": 14431
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011106948620781738,
+      "loss": 0.0061,
+      "step": 14432
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011105076543943218,
+      "loss": 0.0055,
+      "step": 14433
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001110320453215727,
+      "loss": 0.0153,
+      "step": 14434
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011101332585455157,
+      "loss": 0.0024,
+      "step": 14435
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011099460703868154,
+      "loss": 0.0181,
+      "step": 14436
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011097588887427514,
+      "loss": 0.006,
+      "step": 14437
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011095717136164513,
+      "loss": 0.0031,
+      "step": 14438
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011093845450110403,
+      "loss": 0.0155,
+      "step": 14439
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011091973829296446,
+      "loss": 0.0061,
+      "step": 14440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011090102273753904,
+      "loss": 0.0161,
+      "step": 14441
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011088230783514031,
+      "loss": 0.0135,
+      "step": 14442
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011086359358608095,
+      "loss": 0.02,
+      "step": 14443
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011084487999067343,
+      "loss": 0.0131,
+      "step": 14444
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011082616704923028,
+      "loss": 0.001,
+      "step": 14445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011080745476206414,
+      "loss": 0.0036,
+      "step": 14446
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011078874312948743,
+      "loss": 0.0124,
+      "step": 14447
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001107700321518128,
+      "loss": 0.0082,
+      "step": 14448
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011075132182935262,
+      "loss": 0.0012,
+      "step": 14449
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011073261216241949,
+      "loss": 0.0102,
+      "step": 14450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011071390315132583,
+      "loss": 0.0148,
+      "step": 14451
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011069519479638412,
+      "loss": 0.0093,
+      "step": 14452
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011067648709790694,
+      "loss": 0.0082,
+      "step": 14453
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011065778005620662,
+      "loss": 0.0072,
+      "step": 14454
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011063907367159558,
+      "loss": 0.007,
+      "step": 14455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011062036794438639,
+      "loss": 0.0055,
+      "step": 14456
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011060166287489138,
+      "loss": 0.0108,
+      "step": 14457
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001105829584634229,
+      "loss": 0.0028,
+      "step": 14458
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011056425471029346,
+      "loss": 0.0037,
+      "step": 14459
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001105455516158154,
+      "loss": 0.0025,
+      "step": 14460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011052684918030109,
+      "loss": 0.0043,
+      "step": 14461
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011050814740406297,
+      "loss": 0.007,
+      "step": 14462
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011048944628741327,
+      "loss": 0.0054,
+      "step": 14463
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011047074583066444,
+      "loss": 0.0018,
+      "step": 14464
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001104520460341287,
+      "loss": 0.0272,
+      "step": 14465
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011043334689811854,
+      "loss": 0.0028,
+      "step": 14466
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011041464842294615,
+      "loss": 0.0095,
+      "step": 14467
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001103959506089238,
+      "loss": 0.0048,
+      "step": 14468
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001103772534563639,
+      "loss": 0.0009,
+      "step": 14469
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011035855696557859,
+      "loss": 0.0048,
+      "step": 14470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011033986113688024,
+      "loss": 0.0022,
+      "step": 14471
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011032116597058111,
+      "loss": 0.0188,
+      "step": 14472
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011030247146699332,
+      "loss": 0.0325,
+      "step": 14473
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011028377762642929,
+      "loss": 0.0061,
+      "step": 14474
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011026508444920107,
+      "loss": 0.0201,
+      "step": 14475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000110246391935621,
+      "loss": 0.0099,
+      "step": 14476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011022770008600123,
+      "loss": 0.0123,
+      "step": 14477
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011020900890065387,
+      "loss": 0.0039,
+      "step": 14478
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011019031837989123,
+      "loss": 0.0079,
+      "step": 14479
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011017162852402541,
+      "loss": 0.0212,
+      "step": 14480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011015293933336857,
+      "loss": 0.0073,
+      "step": 14481
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011013425080823287,
+      "loss": 0.0064,
+      "step": 14482
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011011556294893036,
+      "loss": 0.0072,
+      "step": 14483
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011009687575577333,
+      "loss": 0.0043,
+      "step": 14484
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011007818922907379,
+      "loss": 0.0128,
+      "step": 14485
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011005950336914378,
+      "loss": 0.0048,
+      "step": 14486
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011004081817629554,
+      "loss": 0.0087,
+      "step": 14487
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011002213365084099,
+      "loss": 0.0127,
+      "step": 14488
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011000344979309232,
+      "loss": 0.0137,
+      "step": 14489
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010998476660336153,
+      "loss": 0.0159,
+      "step": 14490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010996608408196067,
+      "loss": 0.0048,
+      "step": 14491
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010994740222920176,
+      "loss": 0.0049,
+      "step": 14492
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010992872104539685,
+      "loss": 0.0127,
+      "step": 14493
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010991004053085792,
+      "loss": 0.0076,
+      "step": 14494
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010989136068589704,
+      "loss": 0.0068,
+      "step": 14495
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010987268151082607,
+      "loss": 0.0074,
+      "step": 14496
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010985400300595715,
+      "loss": 0.0113,
+      "step": 14497
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010983532517160207,
+      "loss": 0.0054,
+      "step": 14498
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010981664800807296,
+      "loss": 0.007,
+      "step": 14499
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010979797151568167,
+      "loss": 0.0517,
+      "step": 14500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010977929569474008,
+      "loss": 0.0173,
+      "step": 14501
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001097606205455602,
+      "loss": 0.0021,
+      "step": 14502
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010974194606845394,
+      "loss": 0.005,
+      "step": 14503
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010972327226373314,
+      "loss": 0.0037,
+      "step": 14504
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010970459913170976,
+      "loss": 0.003,
+      "step": 14505
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010968592667269557,
+      "loss": 0.0011,
+      "step": 14506
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010966725488700258,
+      "loss": 0.0048,
+      "step": 14507
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010964858377494256,
+      "loss": 0.0055,
+      "step": 14508
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001096299133368273,
+      "loss": 0.0055,
+      "step": 14509
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010961124357296874,
+      "loss": 0.0092,
+      "step": 14510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010959257448367862,
+      "loss": 0.0087,
+      "step": 14511
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010957390606926878,
+      "loss": 0.0009,
+      "step": 14512
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010955523833005108,
+      "loss": 0.0043,
+      "step": 14513
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010953657126633716,
+      "loss": 0.003,
+      "step": 14514
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010951790487843894,
+      "loss": 0.0025,
+      "step": 14515
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010949923916666807,
+      "loss": 0.0128,
+      "step": 14516
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010948057413133641,
+      "loss": 0.0031,
+      "step": 14517
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010946190977275567,
+      "loss": 0.0034,
+      "step": 14518
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010944324609123746,
+      "loss": 0.0124,
+      "step": 14519
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010942458308709369,
+      "loss": 0.0076,
+      "step": 14520
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001094059207606359,
+      "loss": 0.0119,
+      "step": 14521
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010938725911217593,
+      "loss": 0.0029,
+      "step": 14522
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010936859814202537,
+      "loss": 0.0038,
+      "step": 14523
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010934993785049586,
+      "loss": 0.0302,
+      "step": 14524
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001093312782378992,
+      "loss": 0.0071,
+      "step": 14525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001093126193045469,
+      "loss": 0.0035,
+      "step": 14526
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010929396105075071,
+      "loss": 0.0028,
+      "step": 14527
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010927530347682222,
+      "loss": 0.0127,
+      "step": 14528
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010925664658307297,
+      "loss": 0.0075,
+      "step": 14529
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010923799036981469,
+      "loss": 0.0035,
+      "step": 14530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010921933483735889,
+      "loss": 0.0011,
+      "step": 14531
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010920067998601719,
+      "loss": 0.0087,
+      "step": 14532
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010918202581610113,
+      "loss": 0.0083,
+      "step": 14533
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010916337232792228,
+      "loss": 0.0021,
+      "step": 14534
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001091447195217922,
+      "loss": 0.0151,
+      "step": 14535
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010912606739802246,
+      "loss": 0.0092,
+      "step": 14536
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001091074159569245,
+      "loss": 0.0044,
+      "step": 14537
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010908876519880994,
+      "loss": 0.0128,
+      "step": 14538
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010907011512399015,
+      "loss": 0.0111,
+      "step": 14539
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010905146573277678,
+      "loss": 0.0201,
+      "step": 14540
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010903281702548123,
+      "loss": 0.0288,
+      "step": 14541
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010901416900241487,
+      "loss": 0.0027,
+      "step": 14542
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010899552166388931,
+      "loss": 0.0018,
+      "step": 14543
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010897687501021593,
+      "loss": 0.0051,
+      "step": 14544
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001089582290417062,
+      "loss": 0.004,
+      "step": 14545
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010893958375867151,
+      "loss": 0.014,
+      "step": 14546
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010892093916142325,
+      "loss": 0.0032,
+      "step": 14547
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001089022952502729,
+      "loss": 0.0146,
+      "step": 14548
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001088836520255317,
+      "loss": 0.0086,
+      "step": 14549
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010886500948751124,
+      "loss": 0.0037,
+      "step": 14550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010884636763652274,
+      "loss": 0.0138,
+      "step": 14551
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010882772647287753,
+      "loss": 0.0029,
+      "step": 14552
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010880908599688707,
+      "loss": 0.0049,
+      "step": 14553
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010879044620886259,
+      "loss": 0.0078,
+      "step": 14554
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010877180710911541,
+      "loss": 0.0101,
+      "step": 14555
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010875316869795696,
+      "loss": 0.0038,
+      "step": 14556
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010873453097569835,
+      "loss": 0.016,
+      "step": 14557
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010871589394265106,
+      "loss": 0.0051,
+      "step": 14558
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010869725759912627,
+      "loss": 0.0054,
+      "step": 14559
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010867862194543517,
+      "loss": 0.0016,
+      "step": 14560
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010865998698188915,
+      "loss": 0.0101,
+      "step": 14561
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010864135270879932,
+      "loss": 0.0152,
+      "step": 14562
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010862271912647703,
+      "loss": 0.0068,
+      "step": 14563
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001086040862352334,
+      "loss": 0.005,
+      "step": 14564
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010858545403537965,
+      "loss": 0.0104,
+      "step": 14565
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010856682252722706,
+      "loss": 0.0202,
+      "step": 14566
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010854819171108665,
+      "loss": 0.0043,
+      "step": 14567
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010852956158726977,
+      "loss": 0.0023,
+      "step": 14568
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001085109321560875,
+      "loss": 0.0026,
+      "step": 14569
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010849230341785091,
+      "loss": 0.0066,
+      "step": 14570
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010847367537287127,
+      "loss": 0.0026,
+      "step": 14571
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010845504802145956,
+      "loss": 0.017,
+      "step": 14572
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010843642136392704,
+      "loss": 0.007,
+      "step": 14573
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010841779540058471,
+      "loss": 0.0015,
+      "step": 14574
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001083991701317437,
+      "loss": 0.0091,
+      "step": 14575
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010838054555771505,
+      "loss": 0.0042,
+      "step": 14576
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010836192167880991,
+      "loss": 0.0045,
+      "step": 14577
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001083432984953392,
+      "loss": 0.0167,
+      "step": 14578
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010832467600761409,
+      "loss": 0.0135,
+      "step": 14579
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010830605421594552,
+      "loss": 0.0019,
+      "step": 14580
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010828743312064459,
+      "loss": 0.003,
+      "step": 14581
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010826881272202229,
+      "loss": 0.0131,
+      "step": 14582
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001082501930203895,
+      "loss": 0.0084,
+      "step": 14583
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010823157401605734,
+      "loss": 0.0049,
+      "step": 14584
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010821295570933673,
+      "loss": 0.0047,
+      "step": 14585
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010819433810053864,
+      "loss": 0.0005,
+      "step": 14586
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010817572118997405,
+      "loss": 0.0149,
+      "step": 14587
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010815710497795381,
+      "loss": 0.0051,
+      "step": 14588
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010813848946478896,
+      "loss": 0.0047,
+      "step": 14589
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010811987465079028,
+      "loss": 0.0084,
+      "step": 14590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010810126053626883,
+      "loss": 0.0069,
+      "step": 14591
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001080826471215354,
+      "loss": 0.0105,
+      "step": 14592
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010806403440690083,
+      "loss": 0.0032,
+      "step": 14593
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010804542239267612,
+      "loss": 0.0033,
+      "step": 14594
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010802681107917198,
+      "loss": 0.0041,
+      "step": 14595
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010800820046669935,
+      "loss": 0.0131,
+      "step": 14596
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010798959055556907,
+      "loss": 0.0064,
+      "step": 14597
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010797098134609188,
+      "loss": 0.0146,
+      "step": 14598
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010795237283857868,
+      "loss": 0.005,
+      "step": 14599
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010793376503334017,
+      "loss": 0.0005,
+      "step": 14600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010791515793068726,
+      "loss": 0.0146,
+      "step": 14601
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010789655153093065,
+      "loss": 0.0026,
+      "step": 14602
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010787794583438103,
+      "loss": 0.0048,
+      "step": 14603
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010785934084134931,
+      "loss": 0.0064,
+      "step": 14604
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001078407365521461,
+      "loss": 0.0033,
+      "step": 14605
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010782213296708217,
+      "loss": 0.0039,
+      "step": 14606
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010780353008646823,
+      "loss": 0.0119,
+      "step": 14607
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010778492791061496,
+      "loss": 0.0206,
+      "step": 14608
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010776632643983316,
+      "loss": 0.0078,
+      "step": 14609
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010774772567443341,
+      "loss": 0.0097,
+      "step": 14610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010772912561472633,
+      "loss": 0.0028,
+      "step": 14611
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010771052626102273,
+      "loss": 0.0083,
+      "step": 14612
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001076919276136331,
+      "loss": 0.0042,
+      "step": 14613
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001076733296728682,
+      "loss": 0.0039,
+      "step": 14614
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010765473243903856,
+      "loss": 0.0071,
+      "step": 14615
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010763613591245482,
+      "loss": 0.0089,
+      "step": 14616
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010761754009342758,
+      "loss": 0.0056,
+      "step": 14617
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010759894498226743,
+      "loss": 0.0112,
+      "step": 14618
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001075803505792849,
+      "loss": 0.0067,
+      "step": 14619
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010756175688479065,
+      "loss": 0.0123,
+      "step": 14620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010754316389909513,
+      "loss": 0.0159,
+      "step": 14621
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010752457162250893,
+      "loss": 0.0026,
+      "step": 14622
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010750598005534253,
+      "loss": 0.0093,
+      "step": 14623
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010748738919790652,
+      "loss": 0.0083,
+      "step": 14624
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010746879905051137,
+      "loss": 0.002,
+      "step": 14625
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001074502096134675,
+      "loss": 0.0174,
+      "step": 14626
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010743162088708547,
+      "loss": 0.0166,
+      "step": 14627
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010741303287167576,
+      "loss": 0.0147,
+      "step": 14628
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010739444556754873,
+      "loss": 0.0031,
+      "step": 14629
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010737585897501494,
+      "loss": 0.003,
+      "step": 14630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001073572730943847,
+      "loss": 0.0036,
+      "step": 14631
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010733868792596856,
+      "loss": 0.0044,
+      "step": 14632
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010732010347007686,
+      "loss": 0.0111,
+      "step": 14633
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010730151972701992,
+      "loss": 0.0336,
+      "step": 14634
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010728293669710829,
+      "loss": 0.0009,
+      "step": 14635
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010726435438065216,
+      "loss": 0.0185,
+      "step": 14636
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010724577277796203,
+      "loss": 0.0055,
+      "step": 14637
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010722719188934822,
+      "loss": 0.011,
+      "step": 14638
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000107208611715121,
+      "loss": 0.0029,
+      "step": 14639
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010719003225559079,
+      "loss": 0.0139,
+      "step": 14640
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010717145351106778,
+      "loss": 0.0141,
+      "step": 14641
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010715287548186241,
+      "loss": 0.0049,
+      "step": 14642
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010713429816828491,
+      "loss": 0.0107,
+      "step": 14643
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010711572157064549,
+      "loss": 0.0049,
+      "step": 14644
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010709714568925449,
+      "loss": 0.0029,
+      "step": 14645
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010707857052442212,
+      "loss": 0.009,
+      "step": 14646
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010705999607645868,
+      "loss": 0.0143,
+      "step": 14647
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010704142234567434,
+      "loss": 0.007,
+      "step": 14648
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001070228493323793,
+      "loss": 0.0165,
+      "step": 14649
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010700427703688388,
+      "loss": 0.004,
+      "step": 14650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010698570545949818,
+      "loss": 0.0052,
+      "step": 14651
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010696713460053233,
+      "loss": 0.0134,
+      "step": 14652
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010694856446029663,
+      "loss": 0.0014,
+      "step": 14653
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010692999503910112,
+      "loss": 0.0097,
+      "step": 14654
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010691142633725603,
+      "loss": 0.004,
+      "step": 14655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010689285835507145,
+      "loss": 0.0024,
+      "step": 14656
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010687429109285747,
+      "loss": 0.01,
+      "step": 14657
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010685572455092427,
+      "loss": 0.0064,
+      "step": 14658
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010683715872958187,
+      "loss": 0.0031,
+      "step": 14659
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010681859362914045,
+      "loss": 0.0075,
+      "step": 14660
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010680002924991001,
+      "loss": 0.011,
+      "step": 14661
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010678146559220058,
+      "loss": 0.0049,
+      "step": 14662
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010676290265632232,
+      "loss": 0.0076,
+      "step": 14663
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010674434044258514,
+      "loss": 0.0225,
+      "step": 14664
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001067257789512992,
+      "loss": 0.0157,
+      "step": 14665
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010670721818277443,
+      "loss": 0.0049,
+      "step": 14666
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010668865813732076,
+      "loss": 0.0044,
+      "step": 14667
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010667009881524832,
+      "loss": 0.0012,
+      "step": 14668
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010665154021686699,
+      "loss": 0.0029,
+      "step": 14669
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010663298234248678,
+      "loss": 0.0046,
+      "step": 14670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010661442519241764,
+      "loss": 0.0018,
+      "step": 14671
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010659586876696943,
+      "loss": 0.0173,
+      "step": 14672
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010657731306645222,
+      "loss": 0.0008,
+      "step": 14673
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001065587580911758,
+      "loss": 0.0024,
+      "step": 14674
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010654020384145017,
+      "loss": 0.0018,
+      "step": 14675
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010652165031758516,
+      "loss": 0.0278,
+      "step": 14676
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010650309751989061,
+      "loss": 0.0064,
+      "step": 14677
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010648454544867649,
+      "loss": 0.0023,
+      "step": 14678
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010646599410425257,
+      "loss": 0.0049,
+      "step": 14679
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010644744348692872,
+      "loss": 0.0224,
+      "step": 14680
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001064288935970148,
+      "loss": 0.0123,
+      "step": 14681
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010641034443482056,
+      "loss": 0.014,
+      "step": 14682
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010639179600065592,
+      "loss": 0.0181,
+      "step": 14683
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010637324829483058,
+      "loss": 0.0073,
+      "step": 14684
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001063547013176543,
+      "loss": 0.0148,
+      "step": 14685
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010633615506943697,
+      "loss": 0.0085,
+      "step": 14686
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010631760955048819,
+      "loss": 0.0131,
+      "step": 14687
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010629906476111785,
+      "loss": 0.0071,
+      "step": 14688
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010628052070163558,
+      "loss": 0.0051,
+      "step": 14689
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010626197737235113,
+      "loss": 0.0089,
+      "step": 14690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010624343477357427,
+      "loss": 0.0145,
+      "step": 14691
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010622489290561459,
+      "loss": 0.0103,
+      "step": 14692
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010620635176878189,
+      "loss": 0.0084,
+      "step": 14693
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010618781136338577,
+      "loss": 0.0085,
+      "step": 14694
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010616927168973586,
+      "loss": 0.007,
+      "step": 14695
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010615073274814189,
+      "loss": 0.0107,
+      "step": 14696
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001061321945389134,
+      "loss": 0.0111,
+      "step": 14697
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010611365706236013,
+      "loss": 0.0066,
+      "step": 14698
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010609512031879161,
+      "loss": 0.002,
+      "step": 14699
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010607658430851744,
+      "loss": 0.0061,
+      "step": 14700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010605804903184723,
+      "loss": 0.0043,
+      "step": 14701
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010603951448909059,
+      "loss": 0.0063,
+      "step": 14702
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010602098068055696,
+      "loss": 0.0053,
+      "step": 14703
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010600244760655605,
+      "loss": 0.0071,
+      "step": 14704
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010598391526739724,
+      "loss": 0.0089,
+      "step": 14705
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010596538366339017,
+      "loss": 0.0035,
+      "step": 14706
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010594685279484435,
+      "loss": 0.0019,
+      "step": 14707
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010592832266206916,
+      "loss": 0.0095,
+      "step": 14708
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010590979326537424,
+      "loss": 0.0027,
+      "step": 14709
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010589126460506896,
+      "loss": 0.0076,
+      "step": 14710
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010587273668146284,
+      "loss": 0.0049,
+      "step": 14711
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010585420949486534,
+      "loss": 0.002,
+      "step": 14712
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001058356830455858,
+      "loss": 0.0011,
+      "step": 14713
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010581715733393378,
+      "loss": 0.0095,
+      "step": 14714
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010579863236021861,
+      "loss": 0.01,
+      "step": 14715
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010578010812474973,
+      "loss": 0.011,
+      "step": 14716
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010576158462783654,
+      "loss": 0.0074,
+      "step": 14717
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010574306186978831,
+      "loss": 0.0023,
+      "step": 14718
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010572453985091459,
+      "loss": 0.0131,
+      "step": 14719
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010570601857152453,
+      "loss": 0.0149,
+      "step": 14720
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010568749803192762,
+      "loss": 0.0015,
+      "step": 14721
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010566897823243317,
+      "loss": 0.002,
+      "step": 14722
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010565045917335039,
+      "loss": 0.004,
+      "step": 14723
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010563194085498875,
+      "loss": 0.012,
+      "step": 14724
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010561342327765734,
+      "loss": 0.0012,
+      "step": 14725
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010559490644166564,
+      "loss": 0.0066,
+      "step": 14726
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010557639034732283,
+      "loss": 0.0044,
+      "step": 14727
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010555787499493807,
+      "loss": 0.0021,
+      "step": 14728
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010553936038482077,
+      "loss": 0.0032,
+      "step": 14729
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010552084651728004,
+      "loss": 0.0043,
+      "step": 14730
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010550233339262512,
+      "loss": 0.0143,
+      "step": 14731
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010548382101116524,
+      "loss": 0.0198,
+      "step": 14732
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010546530937320954,
+      "loss": 0.0072,
+      "step": 14733
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010544679847906731,
+      "loss": 0.0055,
+      "step": 14734
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010542828832904763,
+      "loss": 0.0055,
+      "step": 14735
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010540977892345963,
+      "loss": 0.0037,
+      "step": 14736
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010539127026261254,
+      "loss": 0.002,
+      "step": 14737
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010537276234681537,
+      "loss": 0.0088,
+      "step": 14738
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001053542551763774,
+      "loss": 0.0055,
+      "step": 14739
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001053357487516076,
+      "loss": 0.0069,
+      "step": 14740
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001053172430728151,
+      "loss": 0.0009,
+      "step": 14741
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010529873814030897,
+      "loss": 0.0082,
+      "step": 14742
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010528023395439829,
+      "loss": 0.0043,
+      "step": 14743
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010526173051539218,
+      "loss": 0.0052,
+      "step": 14744
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001052432278235996,
+      "loss": 0.002,
+      "step": 14745
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010522472587932955,
+      "loss": 0.006,
+      "step": 14746
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010520622468289116,
+      "loss": 0.0022,
+      "step": 14747
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010518772423459331,
+      "loss": 0.0098,
+      "step": 14748
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010516922453474511,
+      "loss": 0.0042,
+      "step": 14749
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010515072558365546,
+      "loss": 0.0052,
+      "step": 14750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010513222738163333,
+      "loss": 0.0083,
+      "step": 14751
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010511372992898772,
+      "loss": 0.0031,
+      "step": 14752
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010509523322602757,
+      "loss": 0.0095,
+      "step": 14753
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010507673727306173,
+      "loss": 0.0069,
+      "step": 14754
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010505824207039924,
+      "loss": 0.0061,
+      "step": 14755
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010503974761834887,
+      "loss": 0.0037,
+      "step": 14756
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010502125391721967,
+      "loss": 0.0036,
+      "step": 14757
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010500276096732042,
+      "loss": 0.0065,
+      "step": 14758
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010498426876895994,
+      "loss": 0.0061,
+      "step": 14759
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010496577732244718,
+      "loss": 0.0014,
+      "step": 14760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001049472866280909,
+      "loss": 0.0044,
+      "step": 14761
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010492879668620003,
+      "loss": 0.0104,
+      "step": 14762
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010491030749708333,
+      "loss": 0.002,
+      "step": 14763
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010489181906104957,
+      "loss": 0.0061,
+      "step": 14764
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010487333137840763,
+      "loss": 0.0058,
+      "step": 14765
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010485484444946619,
+      "loss": 0.0064,
+      "step": 14766
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010483635827453411,
+      "loss": 0.0122,
+      "step": 14767
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010481787285392012,
+      "loss": 0.0022,
+      "step": 14768
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010479938818793286,
+      "loss": 0.003,
+      "step": 14769
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010478090427688122,
+      "loss": 0.0038,
+      "step": 14770
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010476242112107375,
+      "loss": 0.0032,
+      "step": 14771
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010474393872081931,
+      "loss": 0.0004,
+      "step": 14772
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010472545707642648,
+      "loss": 0.0148,
+      "step": 14773
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010470697618820397,
+      "loss": 0.0045,
+      "step": 14774
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010468849605646051,
+      "loss": 0.001,
+      "step": 14775
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001046700166815047,
+      "loss": 0.0058,
+      "step": 14776
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001046515380636451,
+      "loss": 0.0012,
+      "step": 14777
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001046330602031905,
+      "loss": 0.0021,
+      "step": 14778
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010461458310044934,
+      "loss": 0.0004,
+      "step": 14779
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001045961067557304,
+      "loss": 0.0043,
+      "step": 14780
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010457763116934213,
+      "loss": 0.0078,
+      "step": 14781
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010455915634159318,
+      "loss": 0.0048,
+      "step": 14782
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001045406822727921,
+      "loss": 0.003,
+      "step": 14783
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010452220896324742,
+      "loss": 0.004,
+      "step": 14784
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010450373641326769,
+      "loss": 0.0063,
+      "step": 14785
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010448526462316149,
+      "loss": 0.0026,
+      "step": 14786
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010446679359323721,
+      "loss": 0.0044,
+      "step": 14787
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001044483233238035,
+      "loss": 0.0029,
+      "step": 14788
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010442985381516871,
+      "loss": 0.006,
+      "step": 14789
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010441138506764144,
+      "loss": 0.0031,
+      "step": 14790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001043929170815301,
+      "loss": 0.0066,
+      "step": 14791
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010437444985714306,
+      "loss": 0.0052,
+      "step": 14792
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010435598339478889,
+      "loss": 0.0087,
+      "step": 14793
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010433751769477592,
+      "loss": 0.0055,
+      "step": 14794
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010431905275741263,
+      "loss": 0.0058,
+      "step": 14795
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010430058858300739,
+      "loss": 0.0078,
+      "step": 14796
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010428212517186855,
+      "loss": 0.0126,
+      "step": 14797
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010426366252430457,
+      "loss": 0.0028,
+      "step": 14798
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010424520064062368,
+      "loss": 0.0013,
+      "step": 14799
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010422673952113441,
+      "loss": 0.0027,
+      "step": 14800
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010420827916614496,
+      "loss": 0.0096,
+      "step": 14801
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010418981957596363,
+      "loss": 0.0021,
+      "step": 14802
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010417136075089884,
+      "loss": 0.0004,
+      "step": 14803
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010415290269125882,
+      "loss": 0.0012,
+      "step": 14804
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010413444539735183,
+      "loss": 0.0032,
+      "step": 14805
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010411598886948624,
+      "loss": 0.0079,
+      "step": 14806
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010409753310797018,
+      "loss": 0.0015,
+      "step": 14807
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010407907811311203,
+      "loss": 0.0012,
+      "step": 14808
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010406062388521998,
+      "loss": 0.0019,
+      "step": 14809
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010404217042460213,
+      "loss": 0.0011,
+      "step": 14810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010402371773156686,
+      "loss": 0.0046,
+      "step": 14811
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010400526580642223,
+      "loss": 0.0009,
+      "step": 14812
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010398681464947654,
+      "loss": 0.0043,
+      "step": 14813
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010396836426103788,
+      "loss": 0.0037,
+      "step": 14814
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001039499146414144,
+      "loss": 0.0141,
+      "step": 14815
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010393146579091432,
+      "loss": 0.0052,
+      "step": 14816
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010391301770984567,
+      "loss": 0.0018,
+      "step": 14817
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010389457039851669,
+      "loss": 0.0072,
+      "step": 14818
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010387612385723539,
+      "loss": 0.0052,
+      "step": 14819
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010385767808630983,
+      "loss": 0.0126,
+      "step": 14820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010383923308604822,
+      "loss": 0.0092,
+      "step": 14821
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010382078885675849,
+      "loss": 0.0146,
+      "step": 14822
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001038023453987488,
+      "loss": 0.0083,
+      "step": 14823
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010378390271232712,
+      "loss": 0.0084,
+      "step": 14824
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010376546079780152,
+      "loss": 0.0033,
+      "step": 14825
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010374701965547998,
+      "loss": 0.0237,
+      "step": 14826
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010372857928567057,
+      "loss": 0.0005,
+      "step": 14827
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010371013968868115,
+      "loss": 0.0032,
+      "step": 14828
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010369170086481986,
+      "loss": 0.0047,
+      "step": 14829
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001036732628143945,
+      "loss": 0.0017,
+      "step": 14830
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010365482553771315,
+      "loss": 0.0061,
+      "step": 14831
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010363638903508372,
+      "loss": 0.0086,
+      "step": 14832
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010361795330681404,
+      "loss": 0.0011,
+      "step": 14833
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010359951835321211,
+      "loss": 0.0012,
+      "step": 14834
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010358108417458581,
+      "loss": 0.0023,
+      "step": 14835
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010356265077124306,
+      "loss": 0.0063,
+      "step": 14836
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010354421814349171,
+      "loss": 0.0067,
+      "step": 14837
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010352578629163953,
+      "loss": 0.0033,
+      "step": 14838
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010350735521599451,
+      "loss": 0.0078,
+      "step": 14839
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010348892491686437,
+      "loss": 0.0043,
+      "step": 14840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010347049539455707,
+      "loss": 0.0034,
+      "step": 14841
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010345206664938028,
+      "loss": 0.0026,
+      "step": 14842
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001034336386816418,
+      "loss": 0.0009,
+      "step": 14843
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010341521149164952,
+      "loss": 0.0141,
+      "step": 14844
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010339678507971108,
+      "loss": 0.0077,
+      "step": 14845
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001033783594461343,
+      "loss": 0.0026,
+      "step": 14846
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000103359934591227,
+      "loss": 0.0031,
+      "step": 14847
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010334151051529674,
+      "loss": 0.0045,
+      "step": 14848
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001033230872186514,
+      "loss": 0.0025,
+      "step": 14849
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001033046647015986,
+      "loss": 0.0191,
+      "step": 14850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010328624296444601,
+      "loss": 0.005,
+      "step": 14851
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010326782200750137,
+      "loss": 0.0018,
+      "step": 14852
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010324940183107227,
+      "loss": 0.002,
+      "step": 14853
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010323098243546646,
+      "loss": 0.0075,
+      "step": 14854
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010321256382099152,
+      "loss": 0.0036,
+      "step": 14855
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010319414598795507,
+      "loss": 0.0008,
+      "step": 14856
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010317572893666472,
+      "loss": 0.0024,
+      "step": 14857
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010315731266742806,
+      "loss": 0.0029,
+      "step": 14858
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010313889718055277,
+      "loss": 0.0108,
+      "step": 14859
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010312048247634634,
+      "loss": 0.0061,
+      "step": 14860
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010310206855511629,
+      "loss": 0.0065,
+      "step": 14861
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010308365541717029,
+      "loss": 0.0037,
+      "step": 14862
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010306524306281574,
+      "loss": 0.0001,
+      "step": 14863
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010304683149236026,
+      "loss": 0.0113,
+      "step": 14864
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010302842070611134,
+      "loss": 0.0023,
+      "step": 14865
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010301001070437642,
+      "loss": 0.0004,
+      "step": 14866
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010299160148746304,
+      "loss": 0.0004,
+      "step": 14867
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010297319305567864,
+      "loss": 0.0037,
+      "step": 14868
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001029547854093307,
+      "loss": 0.0006,
+      "step": 14869
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010293637854872669,
+      "loss": 0.0114,
+      "step": 14870
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010291797247417392,
+      "loss": 0.0033,
+      "step": 14871
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010289956718597992,
+      "loss": 0.0006,
+      "step": 14872
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010288116268445203,
+      "loss": 0.0025,
+      "step": 14873
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010286275896989773,
+      "loss": 0.0159,
+      "step": 14874
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010284435604262426,
+      "loss": 0.0008,
+      "step": 14875
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010282595390293909,
+      "loss": 0.0006,
+      "step": 14876
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010280755255114954,
+      "loss": 0.0061,
+      "step": 14877
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010278915198756298,
+      "loss": 0.0028,
+      "step": 14878
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010277075221248661,
+      "loss": 0.0045,
+      "step": 14879
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010275235322622791,
+      "loss": 0.0014,
+      "step": 14880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010273395502909402,
+      "loss": 0.0028,
+      "step": 14881
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001027155576213924,
+      "loss": 0.0009,
+      "step": 14882
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010269716100343018,
+      "loss": 0.0072,
+      "step": 14883
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001026787651755146,
+      "loss": 0.0021,
+      "step": 14884
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010266037013795307,
+      "loss": 0.0029,
+      "step": 14885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010264197589105262,
+      "loss": 0.0007,
+      "step": 14886
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010262358243512064,
+      "loss": 0.0028,
+      "step": 14887
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010260518977046426,
+      "loss": 0.011,
+      "step": 14888
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010258679789739061,
+      "loss": 0.01,
+      "step": 14889
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000102568406816207,
+      "loss": 0.0017,
+      "step": 14890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010255001652722048,
+      "loss": 0.0059,
+      "step": 14891
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001025316270307383,
+      "loss": 0.0225,
+      "step": 14892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010251323832706755,
+      "loss": 0.0061,
+      "step": 14893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001024948504165153,
+      "loss": 0.0049,
+      "step": 14894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010247646329938879,
+      "loss": 0.0104,
+      "step": 14895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000102458076975995,
+      "loss": 0.0017,
+      "step": 14896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001024396914466411,
+      "loss": 0.0042,
+      "step": 14897
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001024213067116341,
+      "loss": 0.006,
+      "step": 14898
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010240292277128107,
+      "loss": 0.0058,
+      "step": 14899
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010238453962588914,
+      "loss": 0.0054,
+      "step": 14900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010236615727576528,
+      "loss": 0.0058,
+      "step": 14901
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010234777572121643,
+      "loss": 0.0045,
+      "step": 14902
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010232939496254976,
+      "loss": 0.0005,
+      "step": 14903
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010231101500007211,
+      "loss": 0.0003,
+      "step": 14904
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010229263583409058,
+      "loss": 0.0051,
+      "step": 14905
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010227425746491205,
+      "loss": 0.0003,
+      "step": 14906
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010225587989284353,
+      "loss": 0.0019,
+      "step": 14907
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010223750311819191,
+      "loss": 0.0074,
+      "step": 14908
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010221912714126417,
+      "loss": 0.0076,
+      "step": 14909
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001022007519623672,
+      "loss": 0.0003,
+      "step": 14910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010218237758180793,
+      "loss": 0.0061,
+      "step": 14911
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010216400399989314,
+      "loss": 0.0035,
+      "step": 14912
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010214563121692986,
+      "loss": 0.0049,
+      "step": 14913
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001021272592332248,
+      "loss": 0.0064,
+      "step": 14914
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010210888804908493,
+      "loss": 0.0002,
+      "step": 14915
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010209051766481705,
+      "loss": 0.0155,
+      "step": 14916
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010207214808072786,
+      "loss": 0.0061,
+      "step": 14917
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010205377929712434,
+      "loss": 0.0063,
+      "step": 14918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001020354113143132,
+      "loss": 0.0131,
+      "step": 14919
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010201704413260123,
+      "loss": 0.0063,
+      "step": 14920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010199867775229523,
+      "loss": 0.0007,
+      "step": 14921
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010198031217370184,
+      "loss": 0.001,
+      "step": 14922
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010196194739712797,
+      "loss": 0.0067,
+      "step": 14923
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010194358342288025,
+      "loss": 0.0005,
+      "step": 14924
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010192522025126532,
+      "loss": 0.0117,
+      "step": 14925
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010190685788259004,
+      "loss": 0.0035,
+      "step": 14926
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010188849631716094,
+      "loss": 0.0042,
+      "step": 14927
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001018701355552848,
+      "loss": 0.0062,
+      "step": 14928
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010185177559726828,
+      "loss": 0.0184,
+      "step": 14929
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010183341644341794,
+      "loss": 0.0034,
+      "step": 14930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010181505809404051,
+      "loss": 0.002,
+      "step": 14931
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010179670054944254,
+      "loss": 0.0039,
+      "step": 14932
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010177834380993068,
+      "loss": 0.0044,
+      "step": 14933
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010175998787581152,
+      "loss": 0.0078,
+      "step": 14934
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010174163274739157,
+      "loss": 0.0177,
+      "step": 14935
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001017232784249775,
+      "loss": 0.0144,
+      "step": 14936
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010170492490887577,
+      "loss": 0.0011,
+      "step": 14937
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010168657219939299,
+      "loss": 0.0047,
+      "step": 14938
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010166822029683563,
+      "loss": 0.0069,
+      "step": 14939
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010164986920151018,
+      "loss": 0.0013,
+      "step": 14940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010163151891372327,
+      "loss": 0.0039,
+      "step": 14941
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010161316943378122,
+      "loss": 0.0086,
+      "step": 14942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010159482076199064,
+      "loss": 0.0046,
+      "step": 14943
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010157647289865794,
+      "loss": 0.0069,
+      "step": 14944
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010155812584408945,
+      "loss": 0.0002,
+      "step": 14945
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010153977959859177,
+      "loss": 0.0018,
+      "step": 14946
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010152143416247121,
+      "loss": 0.0064,
+      "step": 14947
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010150308953603426,
+      "loss": 0.0022,
+      "step": 14948
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010148474571958722,
+      "loss": 0.0039,
+      "step": 14949
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010146640271343651,
+      "loss": 0.0063,
+      "step": 14950
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010144806051788849,
+      "loss": 0.0032,
+      "step": 14951
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010142971913324954,
+      "loss": 0.0061,
+      "step": 14952
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010141137855982591,
+      "loss": 0.0026,
+      "step": 14953
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000101393038797924,
+      "loss": 0.0016,
+      "step": 14954
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010137469984785008,
+      "loss": 0.0047,
+      "step": 14955
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010135636170991049,
+      "loss": 0.0062,
+      "step": 14956
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001013380243844115,
+      "loss": 0.0058,
+      "step": 14957
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010131968787165929,
+      "loss": 0.0024,
+      "step": 14958
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001013013521719602,
+      "loss": 0.0068,
+      "step": 14959
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010128301728562048,
+      "loss": 0.0042,
+      "step": 14960
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010126468321294631,
+      "loss": 0.0137,
+      "step": 14961
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010124634995424395,
+      "loss": 0.0016,
+      "step": 14962
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010122801750981955,
+      "loss": 0.0056,
+      "step": 14963
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010120968587997934,
+      "loss": 0.0029,
+      "step": 14964
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010119135506502944,
+      "loss": 0.0043,
+      "step": 14965
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010117302506527608,
+      "loss": 0.0017,
+      "step": 14966
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010115469588102541,
+      "loss": 0.0047,
+      "step": 14967
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010113636751258341,
+      "loss": 0.0195,
+      "step": 14968
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010111803996025641,
+      "loss": 0.003,
+      "step": 14969
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010109971322435032,
+      "loss": 0.0071,
+      "step": 14970
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010108138730517139,
+      "loss": 0.0094,
+      "step": 14971
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010106306220302564,
+      "loss": 0.004,
+      "step": 14972
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010104473791821907,
+      "loss": 0.004,
+      "step": 14973
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010102641445105786,
+      "loss": 0.0042,
+      "step": 14974
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010100809180184795,
+      "loss": 0.0022,
+      "step": 14975
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010098976997089533,
+      "loss": 0.0091,
+      "step": 14976
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010097144895850613,
+      "loss": 0.0018,
+      "step": 14977
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010095312876498623,
+      "loss": 0.0059,
+      "step": 14978
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001009348093906417,
+      "loss": 0.0089,
+      "step": 14979
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010091649083577847,
+      "loss": 0.0003,
+      "step": 14980
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010089817310070244,
+      "loss": 0.014,
+      "step": 14981
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010087985618571969,
+      "loss": 0.0151,
+      "step": 14982
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010086154009113598,
+      "loss": 0.0098,
+      "step": 14983
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010084322481725738,
+      "loss": 0.003,
+      "step": 14984
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010082491036438972,
+      "loss": 0.001,
+      "step": 14985
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001008065967328388,
+      "loss": 0.0056,
+      "step": 14986
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010078828392291064,
+      "loss": 0.0058,
+      "step": 14987
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010076997193491097,
+      "loss": 0.0105,
+      "step": 14988
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010075166076914576,
+      "loss": 0.0051,
+      "step": 14989
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010073335042592076,
+      "loss": 0.0025,
+      "step": 14990
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010071504090554177,
+      "loss": 0.0096,
+      "step": 14991
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010069673220831468,
+      "loss": 0.0024,
+      "step": 14992
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010067842433454516,
+      "loss": 0.0081,
+      "step": 14993
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010066011728453914,
+      "loss": 0.0081,
+      "step": 14994
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010064181105860228,
+      "loss": 0.0071,
+      "step": 14995
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010062350565704029,
+      "loss": 0.0176,
+      "step": 14996
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010060520108015902,
+      "loss": 0.0028,
+      "step": 14997
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010058689732826413,
+      "loss": 0.0069,
+      "step": 14998
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010056859440166126,
+      "loss": 0.0054,
+      "step": 14999
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010055029230065622,
+      "loss": 0.02,
+      "step": 15000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010053199102555463,
+      "loss": 0.0066,
+      "step": 15001
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010051369057666218,
+      "loss": 0.0036,
+      "step": 15002
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010049539095428452,
+      "loss": 0.0028,
+      "step": 15003
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010047709215872724,
+      "loss": 0.0035,
+      "step": 15004
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010045879419029605,
+      "loss": 0.0061,
+      "step": 15005
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010044049704929644,
+      "loss": 0.0039,
+      "step": 15006
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010042220073603417,
+      "loss": 0.0008,
+      "step": 15007
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010040390525081472,
+      "loss": 0.0055,
+      "step": 15008
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001003856105939436,
+      "loss": 0.0049,
+      "step": 15009
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010036731676572649,
+      "loss": 0.005,
+      "step": 15010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010034902376646884,
+      "loss": 0.0127,
+      "step": 15011
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010033073159647623,
+      "loss": 0.0026,
+      "step": 15012
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001003124402560542,
+      "loss": 0.0062,
+      "step": 15013
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010029414974550816,
+      "loss": 0.0005,
+      "step": 15014
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010027586006514368,
+      "loss": 0.0093,
+      "step": 15015
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010025757121526614,
+      "loss": 0.0035,
+      "step": 15016
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010023928319618115,
+      "loss": 0.0061,
+      "step": 15017
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010022099600819404,
+      "loss": 0.001,
+      "step": 15018
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010020270965161018,
+      "loss": 0.0103,
+      "step": 15019
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010018442412673514,
+      "loss": 0.0012,
+      "step": 15020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001001661394338742,
+      "loss": 0.005,
+      "step": 15021
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010014785557333286,
+      "loss": 0.0096,
+      "step": 15022
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010012957254541641,
+      "loss": 0.0056,
+      "step": 15023
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001001112903504302,
+      "loss": 0.0088,
+      "step": 15024
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010009300898867967,
+      "loss": 0.0043,
+      "step": 15025
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001000747284604701,
+      "loss": 0.0005,
+      "step": 15026
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010005644876610673,
+      "loss": 0.0038,
+      "step": 15027
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010003816990589501,
+      "loss": 0.0117,
+      "step": 15028
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001000198918801401,
+      "loss": 0.0019,
+      "step": 15029
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010000161468914741,
+      "loss": 0.001,
+      "step": 15030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.99833383332221e-05,
+      "loss": 0.0035,
+      "step": 15031
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.996506281266944e-05,
+      "loss": 0.0088,
+      "step": 15032
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.99467881277947e-05,
+      "loss": 0.0088,
+      "step": 15033
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.992851427890303e-05,
+      "loss": 0.0023,
+      "step": 15034
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.991024126629976e-05,
+      "loss": 0.0037,
+      "step": 15035
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.989196909029e-05,
+      "loss": 0.0024,
+      "step": 15036
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.987369775117891e-05,
+      "loss": 0.0037,
+      "step": 15037
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.985542724927171e-05,
+      "loss": 0.0021,
+      "step": 15038
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.983715758487348e-05,
+      "loss": 0.0018,
+      "step": 15039
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.981888875828948e-05,
+      "loss": 0.0159,
+      "step": 15040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.980062076982477e-05,
+      "loss": 0.0039,
+      "step": 15041
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.978235361978438e-05,
+      "loss": 0.0044,
+      "step": 15042
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.976408730847351e-05,
+      "loss": 0.0037,
+      "step": 15043
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.974582183619719e-05,
+      "loss": 0.0004,
+      "step": 15044
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.972755720326052e-05,
+      "loss": 0.0052,
+      "step": 15045
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.970929340996858e-05,
+      "loss": 0.0059,
+      "step": 15046
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.96910304566263e-05,
+      "loss": 0.0044,
+      "step": 15047
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.967276834353885e-05,
+      "loss": 0.007,
+      "step": 15048
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.965450707101115e-05,
+      "loss": 0.0039,
+      "step": 15049
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.963624663934817e-05,
+      "loss": 0.0078,
+      "step": 15050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.961798704885497e-05,
+      "loss": 0.006,
+      "step": 15051
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.959972829983645e-05,
+      "loss": 0.0084,
+      "step": 15052
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.958147039259763e-05,
+      "loss": 0.0061,
+      "step": 15053
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.956321332744347e-05,
+      "loss": 0.0047,
+      "step": 15054
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.954495710467879e-05,
+      "loss": 0.0006,
+      "step": 15055
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.952670172460862e-05,
+      "loss": 0.0004,
+      "step": 15056
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.950844718753774e-05,
+      "loss": 0.0028,
+      "step": 15057
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.949019349377116e-05,
+      "loss": 0.0012,
+      "step": 15058
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.947194064361372e-05,
+      "loss": 0.0005,
+      "step": 15059
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.945368863737018e-05,
+      "loss": 0.0047,
+      "step": 15060
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.94354374753455e-05,
+      "loss": 0.001,
+      "step": 15061
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.94171871578444e-05,
+      "loss": 0.0074,
+      "step": 15062
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.939893768517183e-05,
+      "loss": 0.0026,
+      "step": 15063
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.938068905763247e-05,
+      "loss": 0.0186,
+      "step": 15064
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.936244127553113e-05,
+      "loss": 0.0067,
+      "step": 15065
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.934419433917267e-05,
+      "loss": 0.002,
+      "step": 15066
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.932594824886173e-05,
+      "loss": 0.0009,
+      "step": 15067
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.930770300490315e-05,
+      "loss": 0.0078,
+      "step": 15068
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.928945860760163e-05,
+      "loss": 0.0073,
+      "step": 15069
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.927121505726183e-05,
+      "loss": 0.0076,
+      "step": 15070
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.925297235418853e-05,
+      "loss": 0.0065,
+      "step": 15071
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.923473049868635e-05,
+      "loss": 0.0034,
+      "step": 15072
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.921648949106001e-05,
+      "loss": 0.0066,
+      "step": 15073
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.919824933161416e-05,
+      "loss": 0.003,
+      "step": 15074
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.918001002065345e-05,
+      "loss": 0.0044,
+      "step": 15075
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.916177155848249e-05,
+      "loss": 0.006,
+      "step": 15076
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.914353394540594e-05,
+      "loss": 0.0125,
+      "step": 15077
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.91252971817283e-05,
+      "loss": 0.0117,
+      "step": 15078
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.910706126775432e-05,
+      "loss": 0.0055,
+      "step": 15079
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.908882620378839e-05,
+      "loss": 0.002,
+      "step": 15080
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.907059199013526e-05,
+      "loss": 0.005,
+      "step": 15081
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.905235862709934e-05,
+      "loss": 0.0024,
+      "step": 15082
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.903412611498516e-05,
+      "loss": 0.0058,
+      "step": 15083
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.90158944540973e-05,
+      "loss": 0.0062,
+      "step": 15084
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.899766364474025e-05,
+      "loss": 0.0037,
+      "step": 15085
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.897943368721849e-05,
+      "loss": 0.0148,
+      "step": 15086
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.896120458183653e-05,
+      "loss": 0.0013,
+      "step": 15087
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.894297632889872e-05,
+      "loss": 0.014,
+      "step": 15088
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.892474892870966e-05,
+      "loss": 0.0061,
+      "step": 15089
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.890652238157364e-05,
+      "loss": 0.0022,
+      "step": 15090
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.888829668779521e-05,
+      "loss": 0.0048,
+      "step": 15091
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.88700718476787e-05,
+      "loss": 0.004,
+      "step": 15092
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.885184786152844e-05,
+      "loss": 0.0142,
+      "step": 15093
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.883362472964893e-05,
+      "loss": 0.0111,
+      "step": 15094
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.881540245234442e-05,
+      "loss": 0.0046,
+      "step": 15095
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.879718102991932e-05,
+      "loss": 0.0056,
+      "step": 15096
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.877896046267799e-05,
+      "loss": 0.0098,
+      "step": 15097
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.876074075092464e-05,
+      "loss": 0.0043,
+      "step": 15098
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.87425218949637e-05,
+      "loss": 0.0036,
+      "step": 15099
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.87243038950994e-05,
+      "loss": 0.0089,
+      "step": 15100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.870608675163596e-05,
+      "loss": 0.0086,
+      "step": 15101
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.868787046487778e-05,
+      "loss": 0.0089,
+      "step": 15102
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.866965503512891e-05,
+      "loss": 0.0086,
+      "step": 15103
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.865144046269377e-05,
+      "loss": 0.0083,
+      "step": 15104
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.86332267478765e-05,
+      "loss": 0.007,
+      "step": 15105
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.861501389098125e-05,
+      "loss": 0.0126,
+      "step": 15106
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.859680189231233e-05,
+      "loss": 0.0116,
+      "step": 15107
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.857859075217379e-05,
+      "loss": 0.0023,
+      "step": 15108
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.85603804708699e-05,
+      "loss": 0.003,
+      "step": 15109
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.854217104870477e-05,
+      "loss": 0.0113,
+      "step": 15110
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.852396248598247e-05,
+      "loss": 0.0095,
+      "step": 15111
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.850575478300719e-05,
+      "loss": 0.0214,
+      "step": 15112
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.848754794008295e-05,
+      "loss": 0.0071,
+      "step": 15113
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.846934195751396e-05,
+      "loss": 0.0009,
+      "step": 15114
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.845113683560423e-05,
+      "loss": 0.0086,
+      "step": 15115
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.843293257465779e-05,
+      "loss": 0.0017,
+      "step": 15116
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.841472917497874e-05,
+      "loss": 0.0044,
+      "step": 15117
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.839652663687103e-05,
+      "loss": 0.0073,
+      "step": 15118
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.83783249606388e-05,
+      "loss": 0.0016,
+      "step": 15119
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.8360124146586e-05,
+      "loss": 0.009,
+      "step": 15120
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.834192419501654e-05,
+      "loss": 0.0032,
+      "step": 15121
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.83237251062345e-05,
+      "loss": 0.0021,
+      "step": 15122
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.830552688054382e-05,
+      "loss": 0.0158,
+      "step": 15123
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.828732951824836e-05,
+      "loss": 0.0068,
+      "step": 15124
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.826913301965211e-05,
+      "loss": 0.0064,
+      "step": 15125
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.825093738505902e-05,
+      "loss": 0.0135,
+      "step": 15126
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.823274261477296e-05,
+      "loss": 0.0033,
+      "step": 15127
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.821454870909782e-05,
+      "loss": 0.0008,
+      "step": 15128
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.819635566833743e-05,
+      "loss": 0.0057,
+      "step": 15129
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.817816349279573e-05,
+      "loss": 0.0076,
+      "step": 15130
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.815997218277648e-05,
+      "loss": 0.0022,
+      "step": 15131
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.81417817385836e-05,
+      "loss": 0.0017,
+      "step": 15132
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.812359216052085e-05,
+      "loss": 0.0005,
+      "step": 15133
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.810540344889197e-05,
+      "loss": 0.0004,
+      "step": 15134
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.808721560400088e-05,
+      "loss": 0.0011,
+      "step": 15135
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.806902862615123e-05,
+      "loss": 0.0052,
+      "step": 15136
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.805084251564683e-05,
+      "loss": 0.0112,
+      "step": 15137
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.803265727279149e-05,
+      "loss": 0.0032,
+      "step": 15138
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.801447289788878e-05,
+      "loss": 0.0016,
+      "step": 15139
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.799628939124257e-05,
+      "loss": 0.0012,
+      "step": 15140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.797810675315642e-05,
+      "loss": 0.0038,
+      "step": 15141
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.795992498393416e-05,
+      "loss": 0.0031,
+      "step": 15142
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.79417440838794e-05,
+      "loss": 0.0033,
+      "step": 15143
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.79235640532957e-05,
+      "loss": 0.0022,
+      "step": 15144
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.790538489248685e-05,
+      "loss": 0.0033,
+      "step": 15145
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.788720660175637e-05,
+      "loss": 0.0055,
+      "step": 15146
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.786902918140792e-05,
+      "loss": 0.0025,
+      "step": 15147
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.785085263174508e-05,
+      "loss": 0.0065,
+      "step": 15148
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.78326769530714e-05,
+      "loss": 0.0003,
+      "step": 15149
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.781450214569057e-05,
+      "loss": 0.0052,
+      "step": 15150
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.779632820990604e-05,
+      "loss": 0.0004,
+      "step": 15151
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.777815514602131e-05,
+      "loss": 0.0051,
+      "step": 15152
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.775998295434003e-05,
+      "loss": 0.0085,
+      "step": 15153
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.77418116351656e-05,
+      "loss": 0.0016,
+      "step": 15154
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.772364118880161e-05,
+      "loss": 0.0026,
+      "step": 15155
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.770547161555146e-05,
+      "loss": 0.0024,
+      "step": 15156
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.768730291571865e-05,
+      "loss": 0.0049,
+      "step": 15157
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.766913508960662e-05,
+      "loss": 0.0099,
+      "step": 15158
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.765096813751878e-05,
+      "loss": 0.0048,
+      "step": 15159
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.763280205975866e-05,
+      "loss": 0.012,
+      "step": 15160
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.76146368566296e-05,
+      "loss": 0.0063,
+      "step": 15161
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.759647252843492e-05,
+      "loss": 0.0035,
+      "step": 15162
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.757830907547814e-05,
+      "loss": 0.003,
+      "step": 15163
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.756014649806247e-05,
+      "loss": 0.0122,
+      "step": 15164
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.75419847964914e-05,
+      "loss": 0.0019,
+      "step": 15165
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.752382397106821e-05,
+      "loss": 0.0032,
+      "step": 15166
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.750566402209619e-05,
+      "loss": 0.0028,
+      "step": 15167
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.748750494987867e-05,
+      "loss": 0.0072,
+      "step": 15168
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.746934675471894e-05,
+      "loss": 0.0063,
+      "step": 15169
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.745118943692028e-05,
+      "loss": 0.0144,
+      "step": 15170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.743303299678598e-05,
+      "loss": 0.0115,
+      "step": 15171
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.741487743461921e-05,
+      "loss": 0.0146,
+      "step": 15172
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.73967227507233e-05,
+      "loss": 0.0064,
+      "step": 15173
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.73785689454014e-05,
+      "loss": 0.0051,
+      "step": 15174
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.736041601895667e-05,
+      "loss": 0.0032,
+      "step": 15175
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.734226397169245e-05,
+      "loss": 0.0031,
+      "step": 15176
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.732411280391174e-05,
+      "loss": 0.0078,
+      "step": 15177
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.73059625159178e-05,
+      "loss": 0.0026,
+      "step": 15178
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.72878131080138e-05,
+      "loss": 0.0027,
+      "step": 15179
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.726966458050274e-05,
+      "loss": 0.0005,
+      "step": 15180
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.725151693368791e-05,
+      "loss": 0.0061,
+      "step": 15181
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.723337016787222e-05,
+      "loss": 0.002,
+      "step": 15182
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.721522428335893e-05,
+      "loss": 0.0162,
+      "step": 15183
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.719707928045103e-05,
+      "loss": 0.0013,
+      "step": 15184
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.717893515945151e-05,
+      "loss": 0.021,
+      "step": 15185
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.716079192066356e-05,
+      "loss": 0.0044,
+      "step": 15186
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.714264956439005e-05,
+      "loss": 0.0002,
+      "step": 15187
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.712450809093414e-05,
+      "loss": 0.0021,
+      "step": 15188
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.710636750059872e-05,
+      "loss": 0.0013,
+      "step": 15189
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.708822779368678e-05,
+      "loss": 0.0054,
+      "step": 15190
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.707008897050135e-05,
+      "loss": 0.0047,
+      "step": 15191
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.705195103134531e-05,
+      "loss": 0.0008,
+      "step": 15192
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.70338139765217e-05,
+      "loss": 0.002,
+      "step": 15193
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.701567780633336e-05,
+      "loss": 0.0041,
+      "step": 15194
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.699754252108317e-05,
+      "loss": 0.0071,
+      "step": 15195
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.697940812107412e-05,
+      "loss": 0.0026,
+      "step": 15196
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.6961274606609e-05,
+      "loss": 0.0053,
+      "step": 15197
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.694314197799072e-05,
+      "loss": 0.0093,
+      "step": 15198
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.692501023552212e-05,
+      "loss": 0.0003,
+      "step": 15199
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.690687937950603e-05,
+      "loss": 0.0069,
+      "step": 15200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.68887494102453e-05,
+      "loss": 0.0031,
+      "step": 15201
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.687062032804272e-05,
+      "loss": 0.0007,
+      "step": 15202
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.6852492133201e-05,
+      "loss": 0.0119,
+      "step": 15203
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.683436482602307e-05,
+      "loss": 0.0004,
+      "step": 15204
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.681623840681152e-05,
+      "loss": 0.0025,
+      "step": 15205
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.679811287586926e-05,
+      "loss": 0.0037,
+      "step": 15206
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.677998823349892e-05,
+      "loss": 0.0156,
+      "step": 15207
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.676186448000318e-05,
+      "loss": 0.019,
+      "step": 15208
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.674374161568484e-05,
+      "loss": 0.0008,
+      "step": 15209
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.672561964084653e-05,
+      "loss": 0.0019,
+      "step": 15210
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.670749855579095e-05,
+      "loss": 0.0138,
+      "step": 15211
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.668937836082076e-05,
+      "loss": 0.0004,
+      "step": 15212
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.667125905623853e-05,
+      "loss": 0.0044,
+      "step": 15213
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.6653140642347e-05,
+      "loss": 0.0059,
+      "step": 15214
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.663502311944864e-05,
+      "loss": 0.0044,
+      "step": 15215
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.66169064878462e-05,
+      "loss": 0.0037,
+      "step": 15216
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.65987907478422e-05,
+      "loss": 0.0036,
+      "step": 15217
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.658067589973912e-05,
+      "loss": 0.0066,
+      "step": 15218
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.656256194383962e-05,
+      "loss": 0.0127,
+      "step": 15219
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.654444888044626e-05,
+      "loss": 0.0073,
+      "step": 15220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.652633670986141e-05,
+      "loss": 0.009,
+      "step": 15221
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.650822543238777e-05,
+      "loss": 0.013,
+      "step": 15222
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.649011504832766e-05,
+      "loss": 0.0033,
+      "step": 15223
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.647200555798371e-05,
+      "loss": 0.0046,
+      "step": 15224
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.645389696165829e-05,
+      "loss": 0.0054,
+      "step": 15225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.643578925965383e-05,
+      "loss": 0.0075,
+      "step": 15226
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.641768245227284e-05,
+      "loss": 0.0113,
+      "step": 15227
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.639957653981763e-05,
+      "loss": 0.004,
+      "step": 15228
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.638147152259075e-05,
+      "loss": 0.0022,
+      "step": 15229
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.636336740089447e-05,
+      "loss": 0.0057,
+      "step": 15230
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.634526417503116e-05,
+      "loss": 0.0096,
+      "step": 15231
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.632716184530327e-05,
+      "loss": 0.0018,
+      "step": 15232
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.630906041201307e-05,
+      "loss": 0.0058,
+      "step": 15233
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.629095987546294e-05,
+      "loss": 0.0015,
+      "step": 15234
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.627286023595518e-05,
+      "loss": 0.0017,
+      "step": 15235
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.6254761493792e-05,
+      "loss": 0.0003,
+      "step": 15236
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.623666364927582e-05,
+      "loss": 0.0008,
+      "step": 15237
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.621856670270876e-05,
+      "loss": 0.0068,
+      "step": 15238
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.620047065439325e-05,
+      "loss": 0.002,
+      "step": 15239
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.61823755046314e-05,
+      "loss": 0.0065,
+      "step": 15240
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.616428125372546e-05,
+      "loss": 0.0033,
+      "step": 15241
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.614618790197764e-05,
+      "loss": 0.0011,
+      "step": 15242
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.61280954496901e-05,
+      "loss": 0.005,
+      "step": 15243
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.611000389716513e-05,
+      "loss": 0.0024,
+      "step": 15244
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.609191324470482e-05,
+      "loss": 0.0077,
+      "step": 15245
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.607382349261126e-05,
+      "loss": 0.0033,
+      "step": 15246
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.605573464118668e-05,
+      "loss": 0.0097,
+      "step": 15247
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.603764669073318e-05,
+      "loss": 0.0085,
+      "step": 15248
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.601955964155276e-05,
+      "loss": 0.0051,
+      "step": 15249
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.600147349394763e-05,
+      "loss": 0.0007,
+      "step": 15250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.598338824821983e-05,
+      "loss": 0.0078,
+      "step": 15251
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.596530390467141e-05,
+      "loss": 0.009,
+      "step": 15252
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.594722046360444e-05,
+      "loss": 0.0091,
+      "step": 15253
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.592913792532086e-05,
+      "loss": 0.0033,
+      "step": 15254
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.59110562901228e-05,
+      "loss": 0.0056,
+      "step": 15255
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.589297555831213e-05,
+      "loss": 0.0124,
+      "step": 15256
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.587489573019099e-05,
+      "loss": 0.0026,
+      "step": 15257
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.585681680606124e-05,
+      "loss": 0.0041,
+      "step": 15258
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.58387387862248e-05,
+      "loss": 0.0055,
+      "step": 15259
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.58206616709837e-05,
+      "loss": 0.004,
+      "step": 15260
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.580258546063978e-05,
+      "loss": 0.0044,
+      "step": 15261
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.578451015549501e-05,
+      "loss": 0.0114,
+      "step": 15262
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.57664357558513e-05,
+      "loss": 0.0104,
+      "step": 15263
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.574836226201042e-05,
+      "loss": 0.0043,
+      "step": 15264
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.573028967427435e-05,
+      "loss": 0.0066,
+      "step": 15265
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.571221799294482e-05,
+      "loss": 0.0082,
+      "step": 15266
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.56941472183238e-05,
+      "loss": 0.003,
+      "step": 15267
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.567607735071302e-05,
+      "loss": 0.0016,
+      "step": 15268
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.565800839041423e-05,
+      "loss": 0.0002,
+      "step": 15269
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.563994033772936e-05,
+      "loss": 0.0047,
+      "step": 15270
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.562187319296006e-05,
+      "loss": 0.0023,
+      "step": 15271
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.560380695640807e-05,
+      "loss": 0.0037,
+      "step": 15272
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.558574162837526e-05,
+      "loss": 0.0103,
+      "step": 15273
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.556767720916322e-05,
+      "loss": 0.0044,
+      "step": 15274
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.554961369907378e-05,
+      "loss": 0.0016,
+      "step": 15275
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.553155109840857e-05,
+      "loss": 0.0036,
+      "step": 15276
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.551348940746921e-05,
+      "loss": 0.0108,
+      "step": 15277
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.549542862655751e-05,
+      "loss": 0.0039,
+      "step": 15278
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.547736875597495e-05,
+      "loss": 0.0116,
+      "step": 15279
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.545930979602333e-05,
+      "loss": 0.0019,
+      "step": 15280
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.544125174700414e-05,
+      "loss": 0.0035,
+      "step": 15281
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.542319460921904e-05,
+      "loss": 0.0053,
+      "step": 15282
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.540513838296962e-05,
+      "loss": 0.0029,
+      "step": 15283
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.538708306855739e-05,
+      "loss": 0.0027,
+      "step": 15284
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.536902866628406e-05,
+      "loss": 0.0019,
+      "step": 15285
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.535097517645104e-05,
+      "loss": 0.0049,
+      "step": 15286
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.533292259935984e-05,
+      "loss": 0.0022,
+      "step": 15287
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.53148709353121e-05,
+      "loss": 0.0086,
+      "step": 15288
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.529682018460916e-05,
+      "loss": 0.0124,
+      "step": 15289
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.527877034755265e-05,
+      "loss": 0.0048,
+      "step": 15290
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.526072142444394e-05,
+      "loss": 0.0088,
+      "step": 15291
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.524267341558452e-05,
+      "loss": 0.0068,
+      "step": 15292
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.52246263212758e-05,
+      "loss": 0.0092,
+      "step": 15293
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.520658014181925e-05,
+      "loss": 0.01,
+      "step": 15294
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.51885348775162e-05,
+      "loss": 0.0004,
+      "step": 15295
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.517049052866813e-05,
+      "loss": 0.0008,
+      "step": 15296
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.51524470955763e-05,
+      "loss": 0.0084,
+      "step": 15297
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.51344045785422e-05,
+      "loss": 0.0107,
+      "step": 15298
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.511636297786711e-05,
+      "loss": 0.0043,
+      "step": 15299
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.509832229385228e-05,
+      "loss": 0.0054,
+      "step": 15300
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.508028252679919e-05,
+      "loss": 0.0005,
+      "step": 15301
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.506224367700897e-05,
+      "loss": 0.0082,
+      "step": 15302
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.504420574478302e-05,
+      "loss": 0.0009,
+      "step": 15303
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.502616873042261e-05,
+      "loss": 0.0031,
+      "step": 15304
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.500813263422887e-05,
+      "loss": 0.0032,
+      "step": 15305
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.49900974565032e-05,
+      "loss": 0.0171,
+      "step": 15306
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.497206319754667e-05,
+      "loss": 0.0004,
+      "step": 15307
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.495402985766063e-05,
+      "loss": 0.0083,
+      "step": 15308
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.49359974371462e-05,
+      "loss": 0.0088,
+      "step": 15309
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.491796593630448e-05,
+      "loss": 0.0102,
+      "step": 15310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.489993535543677e-05,
+      "loss": 0.0011,
+      "step": 15311
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.488190569484408e-05,
+      "loss": 0.0046,
+      "step": 15312
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.486387695482767e-05,
+      "loss": 0.0006,
+      "step": 15313
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.484584913568857e-05,
+      "loss": 0.0054,
+      "step": 15314
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.482782223772786e-05,
+      "loss": 0.0017,
+      "step": 15315
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.480979626124672e-05,
+      "loss": 0.0032,
+      "step": 15316
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.479177120654611e-05,
+      "loss": 0.0006,
+      "step": 15317
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.47737470739272e-05,
+      "loss": 0.0099,
+      "step": 15318
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.475572386369095e-05,
+      "loss": 0.0036,
+      "step": 15319
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.473770157613834e-05,
+      "loss": 0.0007,
+      "step": 15320
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.471968021157049e-05,
+      "loss": 0.0067,
+      "step": 15321
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.47016597702883e-05,
+      "loss": 0.0057,
+      "step": 15322
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.468364025259275e-05,
+      "loss": 0.0034,
+      "step": 15323
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.466562165878486e-05,
+      "loss": 0.0136,
+      "step": 15324
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.464760398916549e-05,
+      "loss": 0.0081,
+      "step": 15325
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.462958724403568e-05,
+      "loss": 0.0014,
+      "step": 15326
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.46115714236963e-05,
+      "loss": 0.0002,
+      "step": 15327
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.459355652844814e-05,
+      "loss": 0.0045,
+      "step": 15328
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.457554255859226e-05,
+      "loss": 0.0038,
+      "step": 15329
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.455752951442936e-05,
+      "loss": 0.0088,
+      "step": 15330
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.453951739626044e-05,
+      "loss": 0.0015,
+      "step": 15331
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.452150620438627e-05,
+      "loss": 0.0032,
+      "step": 15332
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.450349593910761e-05,
+      "loss": 0.0045,
+      "step": 15333
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.448548660072536e-05,
+      "loss": 0.0171,
+      "step": 15334
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.446747818954025e-05,
+      "loss": 0.0031,
+      "step": 15335
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.444947070585311e-05,
+      "loss": 0.0032,
+      "step": 15336
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.443146414996468e-05,
+      "loss": 0.005,
+      "step": 15337
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.441345852217564e-05,
+      "loss": 0.0033,
+      "step": 15338
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.439545382278682e-05,
+      "loss": 0.0009,
+      "step": 15339
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.437745005209883e-05,
+      "loss": 0.0012,
+      "step": 15340
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.435944721041248e-05,
+      "loss": 0.0061,
+      "step": 15341
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.434144529802839e-05,
+      "loss": 0.0037,
+      "step": 15342
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.432344431524718e-05,
+      "loss": 0.0148,
+      "step": 15343
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.430544426236955e-05,
+      "loss": 0.0066,
+      "step": 15344
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.428744513969618e-05,
+      "loss": 0.003,
+      "step": 15345
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.426944694752756e-05,
+      "loss": 0.0089,
+      "step": 15346
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.425144968616445e-05,
+      "loss": 0.0071,
+      "step": 15347
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.42334533559073e-05,
+      "loss": 0.0064,
+      "step": 15348
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.421545795705682e-05,
+      "loss": 0.0036,
+      "step": 15349
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.419746348991347e-05,
+      "loss": 0.0047,
+      "step": 15350
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.417946995477778e-05,
+      "loss": 0.0013,
+      "step": 15351
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.416147735195035e-05,
+      "loss": 0.0026,
+      "step": 15352
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.41434856817316e-05,
+      "loss": 0.0021,
+      "step": 15353
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.412549494442213e-05,
+      "loss": 0.0154,
+      "step": 15354
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.410750514032234e-05,
+      "loss": 0.0077,
+      "step": 15355
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.408951626973268e-05,
+      "loss": 0.0253,
+      "step": 15356
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.407152833295372e-05,
+      "loss": 0.0014,
+      "step": 15357
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.405354133028572e-05,
+      "loss": 0.0103,
+      "step": 15358
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.403555526202927e-05,
+      "loss": 0.0015,
+      "step": 15359
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.401757012848467e-05,
+      "loss": 0.0017,
+      "step": 15360
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.399958592995228e-05,
+      "loss": 0.0015,
+      "step": 15361
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.398160266673255e-05,
+      "loss": 0.0048,
+      "step": 15362
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.396362033912575e-05,
+      "loss": 0.0024,
+      "step": 15363
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.394563894743234e-05,
+      "loss": 0.0038,
+      "step": 15364
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.392765849195252e-05,
+      "loss": 0.0061,
+      "step": 15365
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.390967897298666e-05,
+      "loss": 0.0035,
+      "step": 15366
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.389170039083502e-05,
+      "loss": 0.0058,
+      "step": 15367
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.387372274579789e-05,
+      "loss": 0.0053,
+      "step": 15368
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.385574603817559e-05,
+      "loss": 0.0141,
+      "step": 15369
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.38377702682683e-05,
+      "loss": 0.0052,
+      "step": 15370
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.381979543637621e-05,
+      "loss": 0.0021,
+      "step": 15371
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.380182154279964e-05,
+      "loss": 0.0044,
+      "step": 15372
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.378384858783874e-05,
+      "loss": 0.0029,
+      "step": 15373
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.37658765717936e-05,
+      "loss": 0.0041,
+      "step": 15374
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.374790549496451e-05,
+      "loss": 0.0044,
+      "step": 15375
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.372993535765159e-05,
+      "loss": 0.0009,
+      "step": 15376
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.371196616015498e-05,
+      "loss": 0.0053,
+      "step": 15377
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.36939979027748e-05,
+      "loss": 0.0059,
+      "step": 15378
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.367603058581107e-05,
+      "loss": 0.0029,
+      "step": 15379
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.365806420956402e-05,
+      "loss": 0.0031,
+      "step": 15380
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.36400987743336e-05,
+      "loss": 0.0055,
+      "step": 15381
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.362213428041997e-05,
+      "loss": 0.007,
+      "step": 15382
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.36041707281231e-05,
+      "loss": 0.0049,
+      "step": 15383
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.358620811774299e-05,
+      "loss": 0.0016,
+      "step": 15384
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.356824644957976e-05,
+      "loss": 0.0049,
+      "step": 15385
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.355028572393327e-05,
+      "loss": 0.0161,
+      "step": 15386
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.35323259411036e-05,
+      "loss": 0.0076,
+      "step": 15387
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.35143671013907e-05,
+      "loss": 0.0049,
+      "step": 15388
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.349640920509444e-05,
+      "loss": 0.0046,
+      "step": 15389
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.347845225251484e-05,
+      "loss": 0.0071,
+      "step": 15390
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.346049624395175e-05,
+      "loss": 0.0077,
+      "step": 15391
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.344254117970515e-05,
+      "loss": 0.0069,
+      "step": 15392
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.342458706007487e-05,
+      "loss": 0.0041,
+      "step": 15393
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.34066338853607e-05,
+      "loss": 0.0021,
+      "step": 15394
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.338868165586268e-05,
+      "loss": 0.004,
+      "step": 15395
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.337073037188046e-05,
+      "loss": 0.0002,
+      "step": 15396
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.335278003371394e-05,
+      "loss": 0.0153,
+      "step": 15397
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.333483064166295e-05,
+      "loss": 0.001,
+      "step": 15398
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.331688219602721e-05,
+      "loss": 0.0134,
+      "step": 15399
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.32989346971066e-05,
+      "loss": 0.0133,
+      "step": 15400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.328098814520081e-05,
+      "loss": 0.0055,
+      "step": 15401
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.326304254060952e-05,
+      "loss": 0.0056,
+      "step": 15402
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.324509788363257e-05,
+      "loss": 0.0069,
+      "step": 15403
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.322715417456958e-05,
+      "loss": 0.0149,
+      "step": 15404
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.32092114137203e-05,
+      "loss": 0.0058,
+      "step": 15405
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.319126960138441e-05,
+      "loss": 0.004,
+      "step": 15406
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.317332873786151e-05,
+      "loss": 0.002,
+      "step": 15407
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.315538882345131e-05,
+      "loss": 0.0019,
+      "step": 15408
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.31374498584534e-05,
+      "loss": 0.0077,
+      "step": 15409
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.311951184316743e-05,
+      "loss": 0.0016,
+      "step": 15410
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.310157477789302e-05,
+      "loss": 0.0038,
+      "step": 15411
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.308363866292965e-05,
+      "loss": 0.0074,
+      "step": 15412
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.3065703498577e-05,
+      "loss": 0.0092,
+      "step": 15413
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.304776928513451e-05,
+      "loss": 0.0089,
+      "step": 15414
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.302983602290186e-05,
+      "loss": 0.0004,
+      "step": 15415
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.301190371217844e-05,
+      "loss": 0.0014,
+      "step": 15416
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.29939723532638e-05,
+      "loss": 0.0121,
+      "step": 15417
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.297604194645743e-05,
+      "loss": 0.0066,
+      "step": 15418
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.295811249205884e-05,
+      "loss": 0.0016,
+      "step": 15419
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.294018399036738e-05,
+      "loss": 0.0043,
+      "step": 15420
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.29222564416826e-05,
+      "loss": 0.0035,
+      "step": 15421
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.290432984630385e-05,
+      "loss": 0.002,
+      "step": 15422
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.288640420453061e-05,
+      "loss": 0.0028,
+      "step": 15423
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.286847951666222e-05,
+      "loss": 0.0024,
+      "step": 15424
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.285055578299801e-05,
+      "loss": 0.002,
+      "step": 15425
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.283263300383745e-05,
+      "loss": 0.0016,
+      "step": 15426
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.281471117947979e-05,
+      "loss": 0.0069,
+      "step": 15427
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.279679031022442e-05,
+      "loss": 0.0061,
+      "step": 15428
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.277887039637066e-05,
+      "loss": 0.0087,
+      "step": 15429
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.276095143821771e-05,
+      "loss": 0.0077,
+      "step": 15430
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.274303343606497e-05,
+      "loss": 0.0036,
+      "step": 15431
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.27251163902116e-05,
+      "loss": 0.0099,
+      "step": 15432
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.270720030095698e-05,
+      "loss": 0.0014,
+      "step": 15433
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.268928516860023e-05,
+      "loss": 0.0036,
+      "step": 15434
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.267137099344058e-05,
+      "loss": 0.0022,
+      "step": 15435
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.265345777577728e-05,
+      "loss": 0.0073,
+      "step": 15436
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.263554551590943e-05,
+      "loss": 0.005,
+      "step": 15437
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.261763421413631e-05,
+      "loss": 0.0044,
+      "step": 15438
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.2599723870757e-05,
+      "loss": 0.0027,
+      "step": 15439
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.258181448607062e-05,
+      "loss": 0.0051,
+      "step": 15440
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.256390606037637e-05,
+      "loss": 0.008,
+      "step": 15441
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.254599859397323e-05,
+      "loss": 0.0013,
+      "step": 15442
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.252809208716047e-05,
+      "loss": 0.0056,
+      "step": 15443
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.251018654023701e-05,
+      "loss": 0.0026,
+      "step": 15444
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.249228195350192e-05,
+      "loss": 0.0011,
+      "step": 15445
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.247437832725435e-05,
+      "loss": 0.0001,
+      "step": 15446
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.24564756617932e-05,
+      "loss": 0.0062,
+      "step": 15447
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.243857395741753e-05,
+      "loss": 0.0058,
+      "step": 15448
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.242067321442634e-05,
+      "loss": 0.0105,
+      "step": 15449
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.240277343311854e-05,
+      "loss": 0.0022,
+      "step": 15450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.238487461379323e-05,
+      "loss": 0.0047,
+      "step": 15451
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.236697675674927e-05,
+      "loss": 0.0032,
+      "step": 15452
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.234907986228552e-05,
+      "loss": 0.005,
+      "step": 15453
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.233118393070103e-05,
+      "loss": 0.0094,
+      "step": 15454
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.231328896229454e-05,
+      "loss": 0.0011,
+      "step": 15455
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.22953949573651e-05,
+      "loss": 0.0091,
+      "step": 15456
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.227750191621146e-05,
+      "loss": 0.0103,
+      "step": 15457
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.225960983913248e-05,
+      "loss": 0.0008,
+      "step": 15458
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.2241718726427e-05,
+      "loss": 0.0049,
+      "step": 15459
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.222382857839387e-05,
+      "loss": 0.0073,
+      "step": 15460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.220593939533186e-05,
+      "loss": 0.0006,
+      "step": 15461
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.218805117753979e-05,
+      "loss": 0.0244,
+      "step": 15462
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.217016392531632e-05,
+      "loss": 0.0089,
+      "step": 15463
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.215227763896033e-05,
+      "loss": 0.002,
+      "step": 15464
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.213439231877046e-05,
+      "loss": 0.0012,
+      "step": 15465
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.211650796504551e-05,
+      "loss": 0.0045,
+      "step": 15466
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.209862457808414e-05,
+      "loss": 0.0043,
+      "step": 15467
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.208074215818498e-05,
+      "loss": 0.0116,
+      "step": 15468
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.206286070564678e-05,
+      "loss": 0.0089,
+      "step": 15469
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.204498022076822e-05,
+      "loss": 0.0055,
+      "step": 15470
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.20271007038478e-05,
+      "loss": 0.0053,
+      "step": 15471
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.200922215518431e-05,
+      "loss": 0.0186,
+      "step": 15472
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.19913445750762e-05,
+      "loss": 0.0079,
+      "step": 15473
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.197346796382222e-05,
+      "loss": 0.0093,
+      "step": 15474
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.195559232172083e-05,
+      "loss": 0.0044,
+      "step": 15475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.193771764907058e-05,
+      "loss": 0.0054,
+      "step": 15476
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.191984394617008e-05,
+      "loss": 0.0093,
+      "step": 15477
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.190197121331778e-05,
+      "loss": 0.0022,
+      "step": 15478
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.188409945081228e-05,
+      "loss": 0.002,
+      "step": 15479
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.186622865895199e-05,
+      "loss": 0.0015,
+      "step": 15480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.18483588380354e-05,
+      "loss": 0.0068,
+      "step": 15481
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.183048998836102e-05,
+      "loss": 0.0071,
+      "step": 15482
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.181262211022722e-05,
+      "loss": 0.0044,
+      "step": 15483
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.179475520393252e-05,
+      "loss": 0.0084,
+      "step": 15484
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.177688926977529e-05,
+      "loss": 0.0016,
+      "step": 15485
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.175902430805385e-05,
+      "loss": 0.0018,
+      "step": 15486
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.174116031906667e-05,
+      "loss": 0.0059,
+      "step": 15487
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.172329730311206e-05,
+      "loss": 0.006,
+      "step": 15488
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.170543526048844e-05,
+      "loss": 0.0092,
+      "step": 15489
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.168757419149405e-05,
+      "loss": 0.0041,
+      "step": 15490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.166971409642727e-05,
+      "loss": 0.0026,
+      "step": 15491
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.165185497558636e-05,
+      "loss": 0.0033,
+      "step": 15492
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.163399682926962e-05,
+      "loss": 0.0099,
+      "step": 15493
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.161613965777528e-05,
+      "loss": 0.0082,
+      "step": 15494
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.159828346140164e-05,
+      "loss": 0.0041,
+      "step": 15495
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.158042824044686e-05,
+      "loss": 0.0028,
+      "step": 15496
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.156257399520928e-05,
+      "loss": 0.0029,
+      "step": 15497
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.154472072598702e-05,
+      "loss": 0.0111,
+      "step": 15498
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.15268684330782e-05,
+      "loss": 0.0009,
+      "step": 15499
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.150901711678106e-05,
+      "loss": 0.0035,
+      "step": 15500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.149116677739377e-05,
+      "loss": 0.007,
+      "step": 15501
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.147331741521443e-05,
+      "loss": 0.0029,
+      "step": 15502
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.145546903054118e-05,
+      "loss": 0.0064,
+      "step": 15503
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.143762162367208e-05,
+      "loss": 0.0012,
+      "step": 15504
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.14197751949053e-05,
+      "loss": 0.001,
+      "step": 15505
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.140192974453876e-05,
+      "loss": 0.0021,
+      "step": 15506
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.138408527287069e-05,
+      "loss": 0.0177,
+      "step": 15507
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.136624178019904e-05,
+      "loss": 0.0022,
+      "step": 15508
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.134839926682179e-05,
+      "loss": 0.0115,
+      "step": 15509
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.133055773303698e-05,
+      "loss": 0.009,
+      "step": 15510
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.131271717914262e-05,
+      "loss": 0.0066,
+      "step": 15511
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.129487760543664e-05,
+      "loss": 0.0018,
+      "step": 15512
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.127703901221707e-05,
+      "loss": 0.0047,
+      "step": 15513
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.125920139978174e-05,
+      "loss": 0.0039,
+      "step": 15514
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.124136476842867e-05,
+      "loss": 0.005,
+      "step": 15515
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.122352911845568e-05,
+      "loss": 0.0098,
+      "step": 15516
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.120569445016075e-05,
+      "loss": 0.0021,
+      "step": 15517
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.118786076384172e-05,
+      "loss": 0.001,
+      "step": 15518
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.117002805979636e-05,
+      "loss": 0.0069,
+      "step": 15519
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.115219633832265e-05,
+      "loss": 0.0119,
+      "step": 15520
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.11343655997183e-05,
+      "loss": 0.0084,
+      "step": 15521
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.111653584428114e-05,
+      "loss": 0.0043,
+      "step": 15522
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.109870707230904e-05,
+      "loss": 0.0028,
+      "step": 15523
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.108087928409965e-05,
+      "loss": 0.0108,
+      "step": 15524
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.106305247995086e-05,
+      "loss": 0.0119,
+      "step": 15525
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.104522666016033e-05,
+      "loss": 0.0086,
+      "step": 15526
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.102740182502576e-05,
+      "loss": 0.0027,
+      "step": 15527
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.100957797484495e-05,
+      "loss": 0.0004,
+      "step": 15528
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.09917551099155e-05,
+      "loss": 0.0076,
+      "step": 15529
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.097393323053517e-05,
+      "loss": 0.0031,
+      "step": 15530
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.095611233700152e-05,
+      "loss": 0.0043,
+      "step": 15531
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.093829242961228e-05,
+      "loss": 0.0064,
+      "step": 15532
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.092047350866503e-05,
+      "loss": 0.0044,
+      "step": 15533
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.090265557445737e-05,
+      "loss": 0.0036,
+      "step": 15534
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.088483862728696e-05,
+      "loss": 0.0004,
+      "step": 15535
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.086702266745135e-05,
+      "loss": 0.0056,
+      "step": 15536
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.084920769524803e-05,
+      "loss": 0.003,
+      "step": 15537
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.083139371097464e-05,
+      "loss": 0.0182,
+      "step": 15538
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.081358071492862e-05,
+      "loss": 0.0089,
+      "step": 15539
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.079576870740754e-05,
+      "loss": 0.0015,
+      "step": 15540
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.07779576887089e-05,
+      "loss": 0.0046,
+      "step": 15541
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.076014765913013e-05,
+      "loss": 0.0069,
+      "step": 15542
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.074233861896874e-05,
+      "loss": 0.005,
+      "step": 15543
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.072453056852215e-05,
+      "loss": 0.0027,
+      "step": 15544
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.070672350808775e-05,
+      "loss": 0.0072,
+      "step": 15545
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.068891743796305e-05,
+      "loss": 0.0055,
+      "step": 15546
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.067111235844531e-05,
+      "loss": 0.0112,
+      "step": 15547
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.065330826983207e-05,
+      "loss": 0.004,
+      "step": 15548
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.06355051724206e-05,
+      "loss": 0.0039,
+      "step": 15549
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.061770306650817e-05,
+      "loss": 0.0027,
+      "step": 15550
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.059990195239227e-05,
+      "loss": 0.0032,
+      "step": 15551
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.058210183037007e-05,
+      "loss": 0.0107,
+      "step": 15552
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.056430270073895e-05,
+      "loss": 0.0045,
+      "step": 15553
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.054650456379623e-05,
+      "loss": 0.003,
+      "step": 15554
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.052870741983901e-05,
+      "loss": 0.0173,
+      "step": 15555
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.051091126916471e-05,
+      "loss": 0.0092,
+      "step": 15556
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.049311611207043e-05,
+      "loss": 0.003,
+      "step": 15557
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.04753219488535e-05,
+      "loss": 0.0026,
+      "step": 15558
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.045752877981105e-05,
+      "loss": 0.0009,
+      "step": 15559
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.043973660524022e-05,
+      "loss": 0.0056,
+      "step": 15560
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.042194542543825e-05,
+      "loss": 0.0141,
+      "step": 15561
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.040415524070221e-05,
+      "loss": 0.0081,
+      "step": 15562
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.038636605132929e-05,
+      "loss": 0.0013,
+      "step": 15563
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.036857785761663e-05,
+      "loss": 0.0025,
+      "step": 15564
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.035079065986122e-05,
+      "loss": 0.006,
+      "step": 15565
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.033300445836025e-05,
+      "loss": 0.0036,
+      "step": 15566
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.031521925341073e-05,
+      "loss": 0.009,
+      "step": 15567
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.029743504530968e-05,
+      "loss": 0.005,
+      "step": 15568
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.02796518343542e-05,
+      "loss": 0.005,
+      "step": 15569
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.026186962084122e-05,
+      "loss": 0.004,
+      "step": 15570
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.024408840506782e-05,
+      "loss": 0.0084,
+      "step": 15571
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.02263081873309e-05,
+      "loss": 0.002,
+      "step": 15572
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.02085289679275e-05,
+      "loss": 0.0117,
+      "step": 15573
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.019075074715452e-05,
+      "loss": 0.0012,
+      "step": 15574
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.017297352530887e-05,
+      "loss": 0.0028,
+      "step": 15575
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.015519730268754e-05,
+      "loss": 0.0012,
+      "step": 15576
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.013742207958741e-05,
+      "loss": 0.0088,
+      "step": 15577
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.011964785630523e-05,
+      "loss": 0.0024,
+      "step": 15578
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.010187463313805e-05,
+      "loss": 0.0082,
+      "step": 15579
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.008410241038257e-05,
+      "loss": 0.0037,
+      "step": 15580
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.006633118833572e-05,
+      "loss": 0.0015,
+      "step": 15581
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.004856096729425e-05,
+      "loss": 0.0034,
+      "step": 15582
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.003079174755499e-05,
+      "loss": 0.005,
+      "step": 15583
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.001302352941471e-05,
+      "loss": 0.0047,
+      "step": 15584
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.999525631317015e-05,
+      "loss": 0.0007,
+      "step": 15585
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.997749009911808e-05,
+      "loss": 0.0061,
+      "step": 15586
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.995972488755527e-05,
+      "loss": 0.0011,
+      "step": 15587
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.99419606787783e-05,
+      "loss": 0.0009,
+      "step": 15588
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.992419747308406e-05,
+      "loss": 0.0007,
+      "step": 15589
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.990643527076902e-05,
+      "loss": 0.0046,
+      "step": 15590
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.988867407213001e-05,
+      "loss": 0.0074,
+      "step": 15591
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.987091387746366e-05,
+      "loss": 0.0074,
+      "step": 15592
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.985315468706643e-05,
+      "loss": 0.0132,
+      "step": 15593
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.983539650123512e-05,
+      "loss": 0.0048,
+      "step": 15594
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.981763932026627e-05,
+      "loss": 0.0042,
+      "step": 15595
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.979988314445638e-05,
+      "loss": 0.0064,
+      "step": 15596
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.978212797410216e-05,
+      "loss": 0.0014,
+      "step": 15597
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.97643738095e-05,
+      "loss": 0.0019,
+      "step": 15598
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.974662065094655e-05,
+      "loss": 0.0066,
+      "step": 15599
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.972886849873828e-05,
+      "loss": 0.0029,
+      "step": 15600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.971111735317162e-05,
+      "loss": 0.0016,
+      "step": 15601
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.969336721454317e-05,
+      "loss": 0.0015,
+      "step": 15602
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.967561808314923e-05,
+      "loss": 0.007,
+      "step": 15603
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.965786995928644e-05,
+      "loss": 0.0082,
+      "step": 15604
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.964012284325108e-05,
+      "loss": 0.003,
+      "step": 15605
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.962237673533957e-05,
+      "loss": 0.001,
+      "step": 15606
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.96046316358484e-05,
+      "loss": 0.003,
+      "step": 15607
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.958688754507383e-05,
+      "loss": 0.0043,
+      "step": 15608
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.956914446331232e-05,
+      "loss": 0.0131,
+      "step": 15609
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.95514023908602e-05,
+      "loss": 0.0014,
+      "step": 15610
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.95336613280137e-05,
+      "loss": 0.0033,
+      "step": 15611
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.951592127506923e-05,
+      "loss": 0.0141,
+      "step": 15612
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.949818223232301e-05,
+      "loss": 0.0024,
+      "step": 15613
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.948044420007138e-05,
+      "loss": 0.0015,
+      "step": 15614
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.946270717861058e-05,
+      "loss": 0.0053,
+      "step": 15615
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.944497116823682e-05,
+      "loss": 0.0018,
+      "step": 15616
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.942723616924635e-05,
+      "loss": 0.0054,
+      "step": 15617
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.94095021819354e-05,
+      "loss": 0.007,
+      "step": 15618
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.939176920660009e-05,
+      "loss": 0.0068,
+      "step": 15619
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.93740372435367e-05,
+      "loss": 0.0012,
+      "step": 15620
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.935630629304125e-05,
+      "loss": 0.0079,
+      "step": 15621
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.933857635541003e-05,
+      "loss": 0.0125,
+      "step": 15622
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.932084743093911e-05,
+      "loss": 0.0079,
+      "step": 15623
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.93031195199245e-05,
+      "loss": 0.0118,
+      "step": 15624
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.928539262266236e-05,
+      "loss": 0.003,
+      "step": 15625
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.926766673944882e-05,
+      "loss": 0.0091,
+      "step": 15626
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.924994187057987e-05,
+      "loss": 0.0218,
+      "step": 15627
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.923221801635161e-05,
+      "loss": 0.0035,
+      "step": 15628
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.921449517705994e-05,
+      "loss": 0.0051,
+      "step": 15629
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.9196773353001e-05,
+      "loss": 0.0096,
+      "step": 15630
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.917905254447067e-05,
+      "loss": 0.0062,
+      "step": 15631
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.916133275176501e-05,
+      "loss": 0.0048,
+      "step": 15632
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.914361397517995e-05,
+      "loss": 0.0002,
+      "step": 15633
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.912589621501135e-05,
+      "loss": 0.0002,
+      "step": 15634
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.910817947155523e-05,
+      "loss": 0.0119,
+      "step": 15635
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.909046374510744e-05,
+      "loss": 0.0082,
+      "step": 15636
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.907274903596387e-05,
+      "loss": 0.0009,
+      "step": 15637
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.905503534442047e-05,
+      "loss": 0.0087,
+      "step": 15638
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.903732267077292e-05,
+      "loss": 0.0092,
+      "step": 15639
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.901961101531724e-05,
+      "loss": 0.0081,
+      "step": 15640
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.900190037834915e-05,
+      "loss": 0.014,
+      "step": 15641
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.89841907601644e-05,
+      "loss": 0.0028,
+      "step": 15642
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.896648216105894e-05,
+      "loss": 0.0033,
+      "step": 15643
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.894877458132833e-05,
+      "loss": 0.0043,
+      "step": 15644
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.893106802126851e-05,
+      "loss": 0.0031,
+      "step": 15645
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.891336248117509e-05,
+      "loss": 0.0013,
+      "step": 15646
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.88956579613438e-05,
+      "loss": 0.004,
+      "step": 15647
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.887795446207044e-05,
+      "loss": 0.0004,
+      "step": 15648
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.886025198365054e-05,
+      "loss": 0.0029,
+      "step": 15649
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.88425505263799e-05,
+      "loss": 0.0011,
+      "step": 15650
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.882485009055412e-05,
+      "loss": 0.0038,
+      "step": 15651
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.880715067646876e-05,
+      "loss": 0.0013,
+      "step": 15652
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.878945228441954e-05,
+      "loss": 0.0017,
+      "step": 15653
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.877175491470197e-05,
+      "loss": 0.0014,
+      "step": 15654
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.87540585676117e-05,
+      "loss": 0.0051,
+      "step": 15655
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.873636324344425e-05,
+      "loss": 0.0012,
+      "step": 15656
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.871866894249516e-05,
+      "loss": 0.0027,
+      "step": 15657
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.870097566505999e-05,
+      "loss": 0.0021,
+      "step": 15658
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.868328341143423e-05,
+      "loss": 0.0043,
+      "step": 15659
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.86655921819134e-05,
+      "loss": 0.0031,
+      "step": 15660
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.864790197679299e-05,
+      "loss": 0.0005,
+      "step": 15661
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.863021279636835e-05,
+      "loss": 0.003,
+      "step": 15662
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.861252464093506e-05,
+      "loss": 0.0051,
+      "step": 15663
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.859483751078842e-05,
+      "loss": 0.0042,
+      "step": 15664
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.857715140622397e-05,
+      "loss": 0.0009,
+      "step": 15665
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.855946632753701e-05,
+      "loss": 0.0007,
+      "step": 15666
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.854178227502294e-05,
+      "loss": 0.004,
+      "step": 15667
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.852409924897711e-05,
+      "loss": 0.0211,
+      "step": 15668
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.85064172496949e-05,
+      "loss": 0.0025,
+      "step": 15669
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.848873627747155e-05,
+      "loss": 0.0033,
+      "step": 15670
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.847105633260247e-05,
+      "loss": 0.0061,
+      "step": 15671
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.845337741538282e-05,
+      "loss": 0.0018,
+      "step": 15672
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.8435699526108e-05,
+      "loss": 0.002,
+      "step": 15673
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.841802266507322e-05,
+      "loss": 0.0059,
+      "step": 15674
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.840034683257364e-05,
+      "loss": 0.0152,
+      "step": 15675
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.838267202890462e-05,
+      "loss": 0.0015,
+      "step": 15676
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.836499825436122e-05,
+      "loss": 0.003,
+      "step": 15677
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.834732550923873e-05,
+      "loss": 0.0072,
+      "step": 15678
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.832965379383229e-05,
+      "loss": 0.0043,
+      "step": 15679
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.831198310843701e-05,
+      "loss": 0.0003,
+      "step": 15680
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.82943134533481e-05,
+      "loss": 0.0015,
+      "step": 15681
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.827664482886058e-05,
+      "loss": 0.0005,
+      "step": 15682
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.825897723526966e-05,
+      "loss": 0.0064,
+      "step": 15683
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.824131067287038e-05,
+      "loss": 0.0084,
+      "step": 15684
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.82236451419577e-05,
+      "loss": 0.005,
+      "step": 15685
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.820598064282683e-05,
+      "loss": 0.01,
+      "step": 15686
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.818831717577267e-05,
+      "loss": 0.0012,
+      "step": 15687
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.817065474109032e-05,
+      "loss": 0.0064,
+      "step": 15688
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.815299333907478e-05,
+      "loss": 0.0052,
+      "step": 15689
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.813533297002094e-05,
+      "loss": 0.0074,
+      "step": 15690
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.811767363422387e-05,
+      "loss": 0.0076,
+      "step": 15691
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.810001533197846e-05,
+      "loss": 0.0053,
+      "step": 15692
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.808235806357956e-05,
+      "loss": 0.001,
+      "step": 15693
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.806470182932224e-05,
+      "loss": 0.0019,
+      "step": 15694
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.804704662950126e-05,
+      "loss": 0.0055,
+      "step": 15695
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.802939246441159e-05,
+      "loss": 0.0019,
+      "step": 15696
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.801173933434801e-05,
+      "loss": 0.0111,
+      "step": 15697
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.79940872396054e-05,
+      "loss": 0.0019,
+      "step": 15698
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.797643618047859e-05,
+      "loss": 0.0068,
+      "step": 15699
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.795878615726234e-05,
+      "loss": 0.0022,
+      "step": 15700
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.79411371702515e-05,
+      "loss": 0.0023,
+      "step": 15701
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.792348921974085e-05,
+      "loss": 0.0014,
+      "step": 15702
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.790584230602504e-05,
+      "loss": 0.0084,
+      "step": 15703
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.788819642939893e-05,
+      "loss": 0.0099,
+      "step": 15704
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.787055159015712e-05,
+      "loss": 0.0068,
+      "step": 15705
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.785290778859446e-05,
+      "loss": 0.0031,
+      "step": 15706
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.783526502500548e-05,
+      "loss": 0.001,
+      "step": 15707
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.781762329968495e-05,
+      "loss": 0.0034,
+      "step": 15708
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.779998261292748e-05,
+      "loss": 0.0065,
+      "step": 15709
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.778234296502771e-05,
+      "loss": 0.0019,
+      "step": 15710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.776470435628026e-05,
+      "loss": 0.0066,
+      "step": 15711
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.774706678697975e-05,
+      "loss": 0.0014,
+      "step": 15712
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.772943025742067e-05,
+      "loss": 0.0021,
+      "step": 15713
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.771179476789772e-05,
+      "loss": 0.0069,
+      "step": 15714
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.769416031870538e-05,
+      "loss": 0.0062,
+      "step": 15715
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.767652691013809e-05,
+      "loss": 0.0011,
+      "step": 15716
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.765889454249053e-05,
+      "loss": 0.0043,
+      "step": 15717
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.764126321605703e-05,
+      "loss": 0.0024,
+      "step": 15718
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.76236329311322e-05,
+      "loss": 0.0065,
+      "step": 15719
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.760600368801044e-05,
+      "loss": 0.0053,
+      "step": 15720
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.758837548698616e-05,
+      "loss": 0.0053,
+      "step": 15721
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.757074832835386e-05,
+      "loss": 0.0072,
+      "step": 15722
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.755312221240788e-05,
+      "loss": 0.0056,
+      "step": 15723
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.753549713944266e-05,
+      "loss": 0.0027,
+      "step": 15724
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.751787310975253e-05,
+      "loss": 0.0077,
+      "step": 15725
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.750025012363185e-05,
+      "loss": 0.0079,
+      "step": 15726
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.748262818137499e-05,
+      "loss": 0.0052,
+      "step": 15727
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.746500728327624e-05,
+      "loss": 0.0025,
+      "step": 15728
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.744738742962991e-05,
+      "loss": 0.0002,
+      "step": 15729
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.742976862073034e-05,
+      "loss": 0.0028,
+      "step": 15730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.741215085687165e-05,
+      "loss": 0.0029,
+      "step": 15731
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.739453413834827e-05,
+      "loss": 0.0003,
+      "step": 15732
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.737691846545426e-05,
+      "loss": 0.0038,
+      "step": 15733
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.735930383848402e-05,
+      "loss": 0.0029,
+      "step": 15734
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.734169025773159e-05,
+      "loss": 0.0044,
+      "step": 15735
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.732407772349116e-05,
+      "loss": 0.0047,
+      "step": 15736
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.730646623605704e-05,
+      "loss": 0.0014,
+      "step": 15737
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.728885579572318e-05,
+      "loss": 0.0072,
+      "step": 15738
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.72712464027839e-05,
+      "loss": 0.0057,
+      "step": 15739
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.725363805753317e-05,
+      "loss": 0.0008,
+      "step": 15740
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.723603076026512e-05,
+      "loss": 0.0062,
+      "step": 15741
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.721842451127383e-05,
+      "loss": 0.0146,
+      "step": 15742
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.720081931085338e-05,
+      "loss": 0.0011,
+      "step": 15743
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.718321515929778e-05,
+      "loss": 0.0042,
+      "step": 15744
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.716561205690105e-05,
+      "loss": 0.0004,
+      "step": 15745
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.71480100039572e-05,
+      "loss": 0.0084,
+      "step": 15746
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.713040900076025e-05,
+      "loss": 0.002,
+      "step": 15747
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.711280904760418e-05,
+      "loss": 0.0139,
+      "step": 15748
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.70952101447828e-05,
+      "loss": 0.001,
+      "step": 15749
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.707761229259027e-05,
+      "loss": 0.008,
+      "step": 15750
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.706001549132027e-05,
+      "loss": 0.0034,
+      "step": 15751
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.704241974126696e-05,
+      "loss": 0.0083,
+      "step": 15752
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.702482504272399e-05,
+      "loss": 0.0036,
+      "step": 15753
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.700723139598532e-05,
+      "loss": 0.0006,
+      "step": 15754
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.698963880134479e-05,
+      "loss": 0.0014,
+      "step": 15755
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.697204725909626e-05,
+      "loss": 0.0018,
+      "step": 15756
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.695445676953349e-05,
+      "loss": 0.0087,
+      "step": 15757
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.693686733295029e-05,
+      "loss": 0.0056,
+      "step": 15758
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.691927894964046e-05,
+      "loss": 0.0067,
+      "step": 15759
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.690169161989775e-05,
+      "loss": 0.0014,
+      "step": 15760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.688410534401589e-05,
+      "loss": 0.0012,
+      "step": 15761
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.686652012228861e-05,
+      "loss": 0.0056,
+      "step": 15762
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.684893595500967e-05,
+      "loss": 0.002,
+      "step": 15763
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.68313528424726e-05,
+      "loss": 0.0041,
+      "step": 15764
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.681377078497129e-05,
+      "loss": 0.0006,
+      "step": 15765
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.679618978279923e-05,
+      "loss": 0.0099,
+      "step": 15766
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.677860983625004e-05,
+      "loss": 0.0036,
+      "step": 15767
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.67610309456175e-05,
+      "loss": 0.0049,
+      "step": 15768
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.674345311119503e-05,
+      "loss": 0.0052,
+      "step": 15769
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.67258763332764e-05,
+      "loss": 0.0042,
+      "step": 15770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.670830061215503e-05,
+      "loss": 0.0021,
+      "step": 15771
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.669072594812449e-05,
+      "loss": 0.0078,
+      "step": 15772
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.667315234147836e-05,
+      "loss": 0.0024,
+      "step": 15773
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.66555797925101e-05,
+      "loss": 0.003,
+      "step": 15774
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.663800830151322e-05,
+      "loss": 0.0036,
+      "step": 15775
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.662043786878123e-05,
+      "loss": 0.0041,
+      "step": 15776
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.660286849460755e-05,
+      "loss": 0.0072,
+      "step": 15777
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.658530017928564e-05,
+      "loss": 0.0085,
+      "step": 15778
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.656773292310894e-05,
+      "loss": 0.0001,
+      "step": 15779
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.655016672637079e-05,
+      "loss": 0.0003,
+      "step": 15780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.653260158936473e-05,
+      "loss": 0.0008,
+      "step": 15781
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.651503751238389e-05,
+      "loss": 0.0007,
+      "step": 15782
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.649747449572188e-05,
+      "loss": 0.0025,
+      "step": 15783
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.64799125396718e-05,
+      "loss": 0.0031,
+      "step": 15784
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.64623516445272e-05,
+      "loss": 0.0063,
+      "step": 15785
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.644479181058119e-05,
+      "loss": 0.0128,
+      "step": 15786
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.642723303812714e-05,
+      "loss": 0.0053,
+      "step": 15787
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.640967532745832e-05,
+      "loss": 0.0099,
+      "step": 15788
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.639211867886792e-05,
+      "loss": 0.0053,
+      "step": 15789
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.637456309264925e-05,
+      "loss": 0.0055,
+      "step": 15790
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.635700856909543e-05,
+      "loss": 0.0078,
+      "step": 15791
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.633945510849972e-05,
+      "loss": 0.0008,
+      "step": 15792
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.632190271115527e-05,
+      "loss": 0.0052,
+      "step": 15793
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.630435137735532e-05,
+      "loss": 0.0008,
+      "step": 15794
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.62868011073928e-05,
+      "loss": 0.0024,
+      "step": 15795
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.626925190156105e-05,
+      "loss": 0.0026,
+      "step": 15796
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.62517037601531e-05,
+      "loss": 0.0028,
+      "step": 15797
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.623415668346202e-05,
+      "loss": 0.0015,
+      "step": 15798
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.621661067178094e-05,
+      "loss": 0.0026,
+      "step": 15799
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.619906572540273e-05,
+      "loss": 0.0027,
+      "step": 15800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.61815218446207e-05,
+      "loss": 0.0044,
+      "step": 15801
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.61639790297276e-05,
+      "loss": 0.0013,
+      "step": 15802
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.614643728101665e-05,
+      "loss": 0.0036,
+      "step": 15803
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.612889659878069e-05,
+      "loss": 0.0033,
+      "step": 15804
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.611135698331272e-05,
+      "loss": 0.0016,
+      "step": 15805
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.609381843490567e-05,
+      "loss": 0.0027,
+      "step": 15806
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.607628095385249e-05,
+      "loss": 0.0031,
+      "step": 15807
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.60587445404461e-05,
+      "loss": 0.003,
+      "step": 15808
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.604120919497936e-05,
+      "loss": 0.0027,
+      "step": 15809
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.602367491774516e-05,
+      "loss": 0.0002,
+      "step": 15810
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.600614170903633e-05,
+      "loss": 0.0006,
+      "step": 15811
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.598860956914576e-05,
+      "loss": 0.0036,
+      "step": 15812
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.597107849836623e-05,
+      "loss": 0.0037,
+      "step": 15813
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.595354849699059e-05,
+      "loss": 0.0066,
+      "step": 15814
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.59360195653115e-05,
+      "loss": 0.0067,
+      "step": 15815
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.591849170362192e-05,
+      "loss": 0.006,
+      "step": 15816
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.590096491221445e-05,
+      "loss": 0.0101,
+      "step": 15817
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.588343919138185e-05,
+      "loss": 0.0003,
+      "step": 15818
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.586591454141682e-05,
+      "loss": 0.0058,
+      "step": 15819
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.584839096261204e-05,
+      "loss": 0.0178,
+      "step": 15820
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.583086845526034e-05,
+      "loss": 0.0012,
+      "step": 15821
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.58133470196542e-05,
+      "loss": 0.0015,
+      "step": 15822
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.579582665608636e-05,
+      "loss": 0.0058,
+      "step": 15823
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.577830736484937e-05,
+      "loss": 0.0027,
+      "step": 15824
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.576078914623589e-05,
+      "loss": 0.0111,
+      "step": 15825
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.574327200053849e-05,
+      "loss": 0.0007,
+      "step": 15826
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.572575592804974e-05,
+      "loss": 0.0067,
+      "step": 15827
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.570824092906217e-05,
+      "loss": 0.0009,
+      "step": 15828
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.569072700386835e-05,
+      "loss": 0.0016,
+      "step": 15829
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.56732141527608e-05,
+      "loss": 0.0002,
+      "step": 15830
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.565570237603199e-05,
+      "loss": 0.0069,
+      "step": 15831
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.563819167397447e-05,
+      "loss": 0.0077,
+      "step": 15832
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.562068204688052e-05,
+      "loss": 0.0054,
+      "step": 15833
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.560317349504281e-05,
+      "loss": 0.0068,
+      "step": 15834
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.558566601875354e-05,
+      "loss": 0.0067,
+      "step": 15835
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.556815961830538e-05,
+      "loss": 0.0101,
+      "step": 15836
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.555065429399052e-05,
+      "loss": 0.0093,
+      "step": 15837
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.553315004610137e-05,
+      "loss": 0.0029,
+      "step": 15838
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.55156468749303e-05,
+      "loss": 0.0096,
+      "step": 15839
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.549814478076966e-05,
+      "loss": 0.0054,
+      "step": 15840
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.548064376391177e-05,
+      "loss": 0.0055,
+      "step": 15841
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.54631438246489e-05,
+      "loss": 0.005,
+      "step": 15842
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.544564496327336e-05,
+      "loss": 0.0054,
+      "step": 15843
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.54281471800774e-05,
+      "loss": 0.0057,
+      "step": 15844
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.541065047535332e-05,
+      "loss": 0.0008,
+      "step": 15845
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.539315484939321e-05,
+      "loss": 0.0027,
+      "step": 15846
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.537566030248945e-05,
+      "loss": 0.0034,
+      "step": 15847
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.535816683493406e-05,
+      "loss": 0.0028,
+      "step": 15848
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.534067444701936e-05,
+      "loss": 0.0057,
+      "step": 15849
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.532318313903747e-05,
+      "loss": 0.0048,
+      "step": 15850
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.530569291128045e-05,
+      "loss": 0.0038,
+      "step": 15851
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.528820376404055e-05,
+      "loss": 0.0023,
+      "step": 15852
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.52707156976097e-05,
+      "loss": 0.0067,
+      "step": 15853
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.52532287122802e-05,
+      "loss": 0.0099,
+      "step": 15854
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.523574280834393e-05,
+      "loss": 0.0056,
+      "step": 15855
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.5218257986093e-05,
+      "loss": 0.0023,
+      "step": 15856
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.520077424581947e-05,
+      "loss": 0.0088,
+      "step": 15857
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.518329158781532e-05,
+      "loss": 0.0068,
+      "step": 15858
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.516581001237255e-05,
+      "loss": 0.0004,
+      "step": 15859
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.514832951978314e-05,
+      "loss": 0.0057,
+      "step": 15860
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.513085011033904e-05,
+      "loss": 0.0072,
+      "step": 15861
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.511337178433219e-05,
+      "loss": 0.0056,
+      "step": 15862
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.509589454205456e-05,
+      "loss": 0.0091,
+      "step": 15863
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.50784183837979e-05,
+      "loss": 0.0181,
+      "step": 15864
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.506094330985432e-05,
+      "loss": 0.011,
+      "step": 15865
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.504346932051545e-05,
+      "loss": 0.0041,
+      "step": 15866
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.502599641607337e-05,
+      "loss": 0.0026,
+      "step": 15867
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.500852459681974e-05,
+      "loss": 0.0099,
+      "step": 15868
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.499105386304644e-05,
+      "loss": 0.0074,
+      "step": 15869
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.497358421504524e-05,
+      "loss": 0.0019,
+      "step": 15870
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.495611565310793e-05,
+      "loss": 0.0118,
+      "step": 15871
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.493864817752627e-05,
+      "loss": 0.0041,
+      "step": 15872
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.4921181788592e-05,
+      "loss": 0.0082,
+      "step": 15873
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.490371648659684e-05,
+      "loss": 0.0016,
+      "step": 15874
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.48862522718325e-05,
+      "loss": 0.0018,
+      "step": 15875
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.486878914459063e-05,
+      "loss": 0.0037,
+      "step": 15876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.485132710516295e-05,
+      "loss": 0.0029,
+      "step": 15877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.483386615384115e-05,
+      "loss": 0.0008,
+      "step": 15878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.481640629091666e-05,
+      "loss": 0.0109,
+      "step": 15879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.479894751668128e-05,
+      "loss": 0.006,
+      "step": 15880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.478148983142657e-05,
+      "loss": 0.0121,
+      "step": 15881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.476403323544408e-05,
+      "loss": 0.0093,
+      "step": 15882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.474657772902545e-05,
+      "loss": 0.0031,
+      "step": 15883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.4729123312462e-05,
+      "loss": 0.006,
+      "step": 15884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.471166998604553e-05,
+      "loss": 0.0086,
+      "step": 15885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.46942177500673e-05,
+      "loss": 0.004,
+      "step": 15886
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.467676660481903e-05,
+      "loss": 0.0012,
+      "step": 15887
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.465931655059199e-05,
+      "loss": 0.0044,
+      "step": 15888
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.46418675876777e-05,
+      "loss": 0.0013,
+      "step": 15889
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.462441971636761e-05,
+      "loss": 0.0044,
+      "step": 15890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.460697293695311e-05,
+      "loss": 0.0046,
+      "step": 15891
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.45895272497256e-05,
+      "loss": 0.0199,
+      "step": 15892
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.457208265497645e-05,
+      "loss": 0.0113,
+      "step": 15893
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.455463915299703e-05,
+      "loss": 0.002,
+      "step": 15894
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.453719674407866e-05,
+      "loss": 0.0019,
+      "step": 15895
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.451975542851275e-05,
+      "loss": 0.0039,
+      "step": 15896
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.45023152065904e-05,
+      "loss": 0.0007,
+      "step": 15897
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.448487607860314e-05,
+      "loss": 0.0138,
+      "step": 15898
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.446743804484199e-05,
+      "loss": 0.0036,
+      "step": 15899
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.445000110559846e-05,
+      "loss": 0.0017,
+      "step": 15900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.443256526116361e-05,
+      "loss": 0.0053,
+      "step": 15901
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.44151305118286e-05,
+      "loss": 0.0027,
+      "step": 15902
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.439769685788482e-05,
+      "loss": 0.0014,
+      "step": 15903
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.438026429962323e-05,
+      "loss": 0.0036,
+      "step": 15904
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.43628328373352e-05,
+      "loss": 0.0025,
+      "step": 15905
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.434540247131171e-05,
+      "loss": 0.0011,
+      "step": 15906
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.432797320184392e-05,
+      "loss": 0.0005,
+      "step": 15907
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.431054502922296e-05,
+      "loss": 0.0049,
+      "step": 15908
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.42931179537399e-05,
+      "loss": 0.0049,
+      "step": 15909
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.427569197568579e-05,
+      "loss": 0.0108,
+      "step": 15910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.425826709535167e-05,
+      "loss": 0.021,
+      "step": 15911
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.424084331302861e-05,
+      "loss": 0.0089,
+      "step": 15912
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.422342062900759e-05,
+      "loss": 0.0126,
+      "step": 15913
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.420599904357967e-05,
+      "loss": 0.002,
+      "step": 15914
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.418857855703564e-05,
+      "loss": 0.0063,
+      "step": 15915
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.417115916966668e-05,
+      "loss": 0.0084,
+      "step": 15916
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.415374088176352e-05,
+      "loss": 0.0039,
+      "step": 15917
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.41363236936173e-05,
+      "loss": 0.0015,
+      "step": 15918
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.411890760551873e-05,
+      "loss": 0.0006,
+      "step": 15919
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.410149261775879e-05,
+      "loss": 0.0014,
+      "step": 15920
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.408407873062826e-05,
+      "loss": 0.0021,
+      "step": 15921
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.406666594441809e-05,
+      "loss": 0.0016,
+      "step": 15922
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.404925425941904e-05,
+      "loss": 0.0005,
+      "step": 15923
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.403184367592194e-05,
+      "loss": 0.008,
+      "step": 15924
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.401443419421757e-05,
+      "loss": 0.0003,
+      "step": 15925
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.39970258145967e-05,
+      "loss": 0.0078,
+      "step": 15926
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.39796185373501e-05,
+      "loss": 0.009,
+      "step": 15927
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.396221236276849e-05,
+      "loss": 0.0035,
+      "step": 15928
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.394480729114265e-05,
+      "loss": 0.0041,
+      "step": 15929
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.39274033227631e-05,
+      "loss": 0.0066,
+      "step": 15930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.391000045792074e-05,
+      "loss": 0.0051,
+      "step": 15931
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.3892598696906e-05,
+      "loss": 0.0111,
+      "step": 15932
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.387519804000974e-05,
+      "loss": 0.0036,
+      "step": 15933
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.385779848752253e-05,
+      "loss": 0.0066,
+      "step": 15934
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.384040003973483e-05,
+      "loss": 0.0213,
+      "step": 15935
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.382300269693744e-05,
+      "loss": 0.0059,
+      "step": 15936
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.380560645942077e-05,
+      "loss": 0.0039,
+      "step": 15937
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.378821132747543e-05,
+      "loss": 0.0054,
+      "step": 15938
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.377081730139192e-05,
+      "loss": 0.0028,
+      "step": 15939
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.375342438146078e-05,
+      "loss": 0.0005,
+      "step": 15940
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.373603256797253e-05,
+      "loss": 0.0033,
+      "step": 15941
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.37186418612176e-05,
+      "loss": 0.0077,
+      "step": 15942
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.370125226148647e-05,
+      "loss": 0.0066,
+      "step": 15943
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.36838637690696e-05,
+      "loss": 0.0026,
+      "step": 15944
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.366647638425735e-05,
+      "loss": 0.0114,
+      "step": 15945
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.364909010734015e-05,
+      "loss": 0.0083,
+      "step": 15946
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.363170493860849e-05,
+      "loss": 0.0018,
+      "step": 15947
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.36143208783525e-05,
+      "loss": 0.0003,
+      "step": 15948
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.359693792686277e-05,
+      "loss": 0.008,
+      "step": 15949
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.35795560844294e-05,
+      "loss": 0.0106,
+      "step": 15950
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.356217535134295e-05,
+      "loss": 0.0014,
+      "step": 15951
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.354479572789354e-05,
+      "loss": 0.0036,
+      "step": 15952
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.352741721437146e-05,
+      "loss": 0.0166,
+      "step": 15953
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.351003981106699e-05,
+      "loss": 0.0072,
+      "step": 15954
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.349266351827034e-05,
+      "loss": 0.0069,
+      "step": 15955
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.347528833627182e-05,
+      "loss": 0.0075,
+      "step": 15956
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.345791426536152e-05,
+      "loss": 0.0105,
+      "step": 15957
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.344054130582966e-05,
+      "loss": 0.0154,
+      "step": 15958
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.342316945796641e-05,
+      "loss": 0.0011,
+      "step": 15959
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.340579872206188e-05,
+      "loss": 0.0036,
+      "step": 15960
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.338842909840623e-05,
+      "loss": 0.0024,
+      "step": 15961
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.337106058728956e-05,
+      "loss": 0.0034,
+      "step": 15962
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.335369318900193e-05,
+      "loss": 0.01,
+      "step": 15963
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.333632690383345e-05,
+      "loss": 0.0069,
+      "step": 15964
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.331896173207418e-05,
+      "loss": 0.0007,
+      "step": 15965
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.330159767401403e-05,
+      "loss": 0.001,
+      "step": 15966
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.328423472994319e-05,
+      "loss": 0.0024,
+      "step": 15967
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.326687290015146e-05,
+      "loss": 0.0081,
+      "step": 15968
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.324951218492905e-05,
+      "loss": 0.0036,
+      "step": 15969
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.323215258456572e-05,
+      "loss": 0.0006,
+      "step": 15970
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.321479409935145e-05,
+      "loss": 0.0102,
+      "step": 15971
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.31974367295762e-05,
+      "loss": 0.0058,
+      "step": 15972
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.318008047552989e-05,
+      "loss": 0.01,
+      "step": 15973
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.316272533750233e-05,
+      "loss": 0.004,
+      "step": 15974
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.314537131578344e-05,
+      "loss": 0.0103,
+      "step": 15975
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.312801841066303e-05,
+      "loss": 0.0128,
+      "step": 15976
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.311066662243096e-05,
+      "loss": 0.0077,
+      "step": 15977
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.309331595137698e-05,
+      "loss": 0.0015,
+      "step": 15978
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.307596639779096e-05,
+      "loss": 0.0034,
+      "step": 15979
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.305861796196268e-05,
+      "loss": 0.0041,
+      "step": 15980
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.304127064418171e-05,
+      "loss": 0.0096,
+      "step": 15981
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.302392444473805e-05,
+      "loss": 0.0033,
+      "step": 15982
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.300657936392116e-05,
+      "loss": 0.0055,
+      "step": 15983
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.298923540202094e-05,
+      "loss": 0.0015,
+      "step": 15984
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.297189255932695e-05,
+      "loss": 0.0018,
+      "step": 15985
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.295455083612882e-05,
+      "loss": 0.0037,
+      "step": 15986
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.293721023271636e-05,
+      "loss": 0.0036,
+      "step": 15987
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.291987074937901e-05,
+      "loss": 0.0029,
+      "step": 15988
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.290253238640646e-05,
+      "loss": 0.0024,
+      "step": 15989
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.288519514408828e-05,
+      "loss": 0.0007,
+      "step": 15990
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.2867859022714e-05,
+      "loss": 0.0035,
+      "step": 15991
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.285052402257321e-05,
+      "loss": 0.005,
+      "step": 15992
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.283319014395543e-05,
+      "loss": 0.0027,
+      "step": 15993
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.281585738715014e-05,
+      "loss": 0.0057,
+      "step": 15994
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.279852575244687e-05,
+      "loss": 0.0092,
+      "step": 15995
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.278119524013507e-05,
+      "loss": 0.0043,
+      "step": 15996
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.276386585050419e-05,
+      "loss": 0.0044,
+      "step": 15997
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.274653758384372e-05,
+      "loss": 0.0023,
+      "step": 15998
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.27292104404429e-05,
+      "loss": 0.0004,
+      "step": 15999
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.27118844205914e-05,
+      "loss": 0.018,
+      "step": 16000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.269455952457831e-05,
+      "loss": 0.0022,
+      "step": 16001
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.267723575269323e-05,
+      "loss": 0.0008,
+      "step": 16002
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.265991310522534e-05,
+      "loss": 0.0034,
+      "step": 16003
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.264259158246402e-05,
+      "loss": 0.0049,
+      "step": 16004
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.262527118469855e-05,
+      "loss": 0.0049,
+      "step": 16005
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.260795191221826e-05,
+      "loss": 0.0133,
+      "step": 16006
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.259063376531236e-05,
+      "loss": 0.0016,
+      "step": 16007
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.257331674427012e-05,
+      "loss": 0.0012,
+      "step": 16008
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.255600084938079e-05,
+      "loss": 0.0058,
+      "step": 16009
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.253868608093354e-05,
+      "loss": 0.0075,
+      "step": 16010
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.252137243921761e-05,
+      "loss": 0.0082,
+      "step": 16011
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.250405992452206e-05,
+      "loss": 0.0051,
+      "step": 16012
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.248674853713621e-05,
+      "loss": 0.0019,
+      "step": 16013
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.246943827734897e-05,
+      "loss": 0.0053,
+      "step": 16014
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.245212914544968e-05,
+      "loss": 0.0114,
+      "step": 16015
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.243482114172739e-05,
+      "loss": 0.0026,
+      "step": 16016
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.241751426647098e-05,
+      "loss": 0.0053,
+      "step": 16017
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.240020851996979e-05,
+      "loss": 0.0008,
+      "step": 16018
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.23829039025126e-05,
+      "loss": 0.0169,
+      "step": 16019
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.236560041438866e-05,
+      "loss": 0.0023,
+      "step": 16020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.234829805588681e-05,
+      "loss": 0.003,
+      "step": 16021
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.233099682729609e-05,
+      "loss": 0.0044,
+      "step": 16022
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.231369672890545e-05,
+      "loss": 0.0036,
+      "step": 16023
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.229639776100386e-05,
+      "loss": 0.0011,
+      "step": 16024
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.227909992388021e-05,
+      "loss": 0.0083,
+      "step": 16025
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.226180321782342e-05,
+      "loss": 0.0062,
+      "step": 16026
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.22445076431224e-05,
+      "loss": 0.0019,
+      "step": 16027
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.222721320006598e-05,
+      "loss": 0.005,
+      "step": 16028
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.220991988894303e-05,
+      "loss": 0.0059,
+      "step": 16029
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.21926277100424e-05,
+      "loss": 0.0111,
+      "step": 16030
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.217533666365292e-05,
+      "loss": 0.0081,
+      "step": 16031
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.215804675006323e-05,
+      "loss": 0.0085,
+      "step": 16032
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.214075796956236e-05,
+      "loss": 0.0015,
+      "step": 16033
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.212347032243878e-05,
+      "loss": 0.01,
+      "step": 16034
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.210618380898151e-05,
+      "loss": 0.0008,
+      "step": 16035
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.208889842947905e-05,
+      "loss": 0.0098,
+      "step": 16036
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.207161418422019e-05,
+      "loss": 0.0029,
+      "step": 16037
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.20543310734936e-05,
+      "loss": 0.0009,
+      "step": 16038
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.203704909758792e-05,
+      "loss": 0.0012,
+      "step": 16039
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.201976825679183e-05,
+      "loss": 0.0011,
+      "step": 16040
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.200248855139393e-05,
+      "loss": 0.0028,
+      "step": 16041
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.198520998168282e-05,
+      "loss": 0.0031,
+      "step": 16042
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.196793254794709e-05,
+      "loss": 0.0099,
+      "step": 16043
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.195065625047536e-05,
+      "loss": 0.0076,
+      "step": 16044
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.1933381089556e-05,
+      "loss": 0.0191,
+      "step": 16045
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.191610706547776e-05,
+      "loss": 0.0159,
+      "step": 16046
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.189883417852901e-05,
+      "loss": 0.0041,
+      "step": 16047
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.188156242899828e-05,
+      "loss": 0.0104,
+      "step": 16048
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.186429181717412e-05,
+      "loss": 0.007,
+      "step": 16049
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.184702234334476e-05,
+      "loss": 0.0039,
+      "step": 16050
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.18297540077989e-05,
+      "loss": 0.0062,
+      "step": 16051
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.181248681082472e-05,
+      "loss": 0.0015,
+      "step": 16052
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.179522075271084e-05,
+      "loss": 0.0078,
+      "step": 16053
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.177795583374545e-05,
+      "loss": 0.0025,
+      "step": 16054
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.176069205421698e-05,
+      "loss": 0.0086,
+      "step": 16055
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.174342941441377e-05,
+      "loss": 0.0036,
+      "step": 16056
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.172616791462412e-05,
+      "loss": 0.0017,
+      "step": 16057
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.170890755513635e-05,
+      "loss": 0.0062,
+      "step": 16058
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.169164833623874e-05,
+      "loss": 0.0042,
+      "step": 16059
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.167439025821954e-05,
+      "loss": 0.0078,
+      "step": 16060
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.165713332136699e-05,
+      "loss": 0.0031,
+      "step": 16061
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.163987752596938e-05,
+      "loss": 0.0006,
+      "step": 16062
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.162262287231473e-05,
+      "loss": 0.0003,
+      "step": 16063
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.16053693606915e-05,
+      "loss": 0.0073,
+      "step": 16064
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.158811699138754e-05,
+      "loss": 0.0029,
+      "step": 16065
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.157086576469131e-05,
+      "loss": 0.0044,
+      "step": 16066
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.155361568089073e-05,
+      "loss": 0.003,
+      "step": 16067
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.153636674027389e-05,
+      "loss": 0.0106,
+      "step": 16068
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.151911894312911e-05,
+      "loss": 0.0068,
+      "step": 16069
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.150187228974418e-05,
+      "loss": 0.0016,
+      "step": 16070
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.148462678040738e-05,
+      "loss": 0.0125,
+      "step": 16071
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.146738241540662e-05,
+      "loss": 0.0028,
+      "step": 16072
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.145013919502992e-05,
+      "loss": 0.0071,
+      "step": 16073
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.14328971195653e-05,
+      "loss": 0.0036,
+      "step": 16074
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.141565618930071e-05,
+      "loss": 0.0052,
+      "step": 16075
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.139841640452414e-05,
+      "loss": 0.0137,
+      "step": 16076
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.138117776552349e-05,
+      "loss": 0.0028,
+      "step": 16077
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.136394027258672e-05,
+      "loss": 0.0053,
+      "step": 16078
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.13467039260017e-05,
+      "loss": 0.0102,
+      "step": 16079
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.132946872605632e-05,
+      "loss": 0.0012,
+      "step": 16080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.131223467303844e-05,
+      "loss": 0.0022,
+      "step": 16081
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.129500176723596e-05,
+      "loss": 0.0025,
+      "step": 16082
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.12777700089365e-05,
+      "loss": 0.0039,
+      "step": 16083
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.126053939842812e-05,
+      "loss": 0.0046,
+      "step": 16084
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.12433099359984e-05,
+      "loss": 0.0175,
+      "step": 16085
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.12260816219353e-05,
+      "loss": 0.006,
+      "step": 16086
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.120885445652639e-05,
+      "loss": 0.0083,
+      "step": 16087
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.119162844005946e-05,
+      "loss": 0.0025,
+      "step": 16088
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.117440357282223e-05,
+      "loss": 0.0024,
+      "step": 16089
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.115717985510238e-05,
+      "loss": 0.0082,
+      "step": 16090
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.113995728718756e-05,
+      "loss": 0.0031,
+      "step": 16091
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.112273586936544e-05,
+      "loss": 0.0078,
+      "step": 16092
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.110551560192365e-05,
+      "loss": 0.0049,
+      "step": 16093
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.10882964851498e-05,
+      "loss": 0.0075,
+      "step": 16094
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.107107851933154e-05,
+      "loss": 0.0029,
+      "step": 16095
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.105386170475627e-05,
+      "loss": 0.0074,
+      "step": 16096
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.103664604171176e-05,
+      "loss": 0.0047,
+      "step": 16097
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.101943153048531e-05,
+      "loss": 0.0042,
+      "step": 16098
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.100221817136464e-05,
+      "loss": 0.0019,
+      "step": 16099
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.098500596463722e-05,
+      "loss": 0.0017,
+      "step": 16100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.096779491059037e-05,
+      "loss": 0.0042,
+      "step": 16101
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.095058500951176e-05,
+      "loss": 0.0049,
+      "step": 16102
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.09333762616886e-05,
+      "loss": 0.0021,
+      "step": 16103
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.091616866740855e-05,
+      "loss": 0.001,
+      "step": 16104
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.089896222695882e-05,
+      "loss": 0.0204,
+      "step": 16105
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.088175694062686e-05,
+      "loss": 0.0044,
+      "step": 16106
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.086455280870005e-05,
+      "loss": 0.0046,
+      "step": 16107
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.08473498314657e-05,
+      "loss": 0.0182,
+      "step": 16108
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.083014800921114e-05,
+      "loss": 0.0091,
+      "step": 16109
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.081294734222367e-05,
+      "loss": 0.0048,
+      "step": 16110
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.079574783079058e-05,
+      "loss": 0.0018,
+      "step": 16111
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.077854947519915e-05,
+      "loss": 0.0051,
+      "step": 16112
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.076135227573665e-05,
+      "loss": 0.0044,
+      "step": 16113
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.074415623269015e-05,
+      "loss": 0.0105,
+      "step": 16114
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.072696134634709e-05,
+      "loss": 0.0005,
+      "step": 16115
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.070976761699443e-05,
+      "loss": 0.008,
+      "step": 16116
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.069257504491957e-05,
+      "loss": 0.0066,
+      "step": 16117
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.067538363040944e-05,
+      "loss": 0.0015,
+      "step": 16118
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.065819337375127e-05,
+      "loss": 0.0055,
+      "step": 16119
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.064100427523218e-05,
+      "loss": 0.0013,
+      "step": 16120
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.062381633513918e-05,
+      "loss": 0.0058,
+      "step": 16121
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.060662955375953e-05,
+      "loss": 0.0101,
+      "step": 16122
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.058944393138007e-05,
+      "loss": 0.0066,
+      "step": 16123
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.057225946828795e-05,
+      "loss": 0.0015,
+      "step": 16124
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.055507616477013e-05,
+      "loss": 0.0036,
+      "step": 16125
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.053789402111364e-05,
+      "loss": 0.0079,
+      "step": 16126
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.052071303760541e-05,
+      "loss": 0.0018,
+      "step": 16127
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.050353321453245e-05,
+      "loss": 0.0034,
+      "step": 16128
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.048635455218166e-05,
+      "loss": 0.0116,
+      "step": 16129
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.046917705083998e-05,
+      "loss": 0.0022,
+      "step": 16130
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.045200071079427e-05,
+      "loss": 0.0055,
+      "step": 16131
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.043482553233144e-05,
+      "loss": 0.007,
+      "step": 16132
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.04176515157384e-05,
+      "loss": 0.0035,
+      "step": 16133
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.040047866130179e-05,
+      "loss": 0.0064,
+      "step": 16134
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.038330696930869e-05,
+      "loss": 0.0016,
+      "step": 16135
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.036613644004574e-05,
+      "loss": 0.0032,
+      "step": 16136
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.034896707379973e-05,
+      "loss": 0.0018,
+      "step": 16137
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.033179887085745e-05,
+      "loss": 0.0022,
+      "step": 16138
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.031463183150563e-05,
+      "loss": 0.0012,
+      "step": 16139
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.029746595603101e-05,
+      "loss": 0.0023,
+      "step": 16140
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.028030124472027e-05,
+      "loss": 0.0088,
+      "step": 16141
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.026313769786011e-05,
+      "loss": 0.0031,
+      "step": 16142
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.02459753157372e-05,
+      "loss": 0.0086,
+      "step": 16143
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.022881409863817e-05,
+      "loss": 0.0055,
+      "step": 16144
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.021165404684963e-05,
+      "loss": 0.0053,
+      "step": 16145
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.019449516065827e-05,
+      "loss": 0.0035,
+      "step": 16146
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.01773374403505e-05,
+      "loss": 0.002,
+      "step": 16147
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.01601808862131e-05,
+      "loss": 0.0047,
+      "step": 16148
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.01430254985324e-05,
+      "loss": 0.0005,
+      "step": 16149
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.012587127759515e-05,
+      "loss": 0.001,
+      "step": 16150
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.010871822368769e-05,
+      "loss": 0.0095,
+      "step": 16151
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.009156633709651e-05,
+      "loss": 0.0045,
+      "step": 16152
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.007441561810825e-05,
+      "loss": 0.011,
+      "step": 16153
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.00572660670091e-05,
+      "loss": 0.0034,
+      "step": 16154
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.00401176840858e-05,
+      "loss": 0.0014,
+      "step": 16155
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.002297046962452e-05,
+      "loss": 0.0048,
+      "step": 16156
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.00058244239117e-05,
+      "loss": 0.0047,
+      "step": 16157
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.998867954723374e-05,
+      "loss": 0.0063,
+      "step": 16158
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.997153583987698e-05,
+      "loss": 0.0029,
+      "step": 16159
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.995439330212779e-05,
+      "loss": 0.0085,
+      "step": 16160
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.993725193427244e-05,
+      "loss": 0.0117,
+      "step": 16161
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.992011173659722e-05,
+      "loss": 0.0008,
+      "step": 16162
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.990297270938844e-05,
+      "loss": 0.0052,
+      "step": 16163
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.988583485293239e-05,
+      "loss": 0.0139,
+      "step": 16164
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.986869816751512e-05,
+      "loss": 0.0027,
+      "step": 16165
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.98515626534231e-05,
+      "loss": 0.007,
+      "step": 16166
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.983442831094226e-05,
+      "loss": 0.0064,
+      "step": 16167
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.981729514035906e-05,
+      "loss": 0.0026,
+      "step": 16168
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.980016314195944e-05,
+      "loss": 0.0032,
+      "step": 16169
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.97830323160296e-05,
+      "loss": 0.0084,
+      "step": 16170
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.976590266285567e-05,
+      "loss": 0.003,
+      "step": 16171
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.974877418272376e-05,
+      "loss": 0.0119,
+      "step": 16172
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.973164687591991e-05,
+      "loss": 0.0044,
+      "step": 16173
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.971452074273019e-05,
+      "loss": 0.0059,
+      "step": 16174
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.969739578344064e-05,
+      "loss": 0.0008,
+      "step": 16175
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.968027199833732e-05,
+      "loss": 0.0005,
+      "step": 16176
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.966314938770617e-05,
+      "loss": 0.0019,
+      "step": 16177
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.96460279518332e-05,
+      "loss": 0.0003,
+      "step": 16178
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.962890769100444e-05,
+      "loss": 0.0028,
+      "step": 16179
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.961178860550563e-05,
+      "loss": 0.0002,
+      "step": 16180
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.959467069562287e-05,
+      "loss": 0.0146,
+      "step": 16181
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.957755396164201e-05,
+      "loss": 0.0026,
+      "step": 16182
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.956043840384893e-05,
+      "loss": 0.0036,
+      "step": 16183
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.954332402252956e-05,
+      "loss": 0.0073,
+      "step": 16184
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.952621081796955e-05,
+      "loss": 0.0236,
+      "step": 16185
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.950909879045497e-05,
+      "loss": 0.003,
+      "step": 16186
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.949198794027144e-05,
+      "loss": 0.002,
+      "step": 16187
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.947487826770478e-05,
+      "loss": 0.0034,
+      "step": 16188
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.945776977304082e-05,
+      "loss": 0.0021,
+      "step": 16189
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.944066245656524e-05,
+      "loss": 0.0082,
+      "step": 16190
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.942355631856381e-05,
+      "loss": 0.0007,
+      "step": 16191
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.940645135932218e-05,
+      "loss": 0.0024,
+      "step": 16192
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.938934757912608e-05,
+      "loss": 0.0009,
+      "step": 16193
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.937224497826117e-05,
+      "loss": 0.0035,
+      "step": 16194
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.93551435570131e-05,
+      "loss": 0.0153,
+      "step": 16195
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.933804331566745e-05,
+      "loss": 0.0075,
+      "step": 16196
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.932094425450993e-05,
+      "loss": 0.0038,
+      "step": 16197
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.930384637382595e-05,
+      "loss": 0.0039,
+      "step": 16198
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.92867496739013e-05,
+      "loss": 0.0029,
+      "step": 16199
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.926965415502128e-05,
+      "loss": 0.0017,
+      "step": 16200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.925255981747168e-05,
+      "loss": 0.0028,
+      "step": 16201
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.923546666153782e-05,
+      "loss": 0.0049,
+      "step": 16202
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.921837468750524e-05,
+      "loss": 0.0047,
+      "step": 16203
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.920128389565941e-05,
+      "loss": 0.003,
+      "step": 16204
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.918419428628571e-05,
+      "loss": 0.006,
+      "step": 16205
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.916710585966978e-05,
+      "loss": 0.0038,
+      "step": 16206
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.915001861609684e-05,
+      "loss": 0.0016,
+      "step": 16207
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.91329325558523e-05,
+      "loss": 0.0082,
+      "step": 16208
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.911584767922155e-05,
+      "loss": 0.0047,
+      "step": 16209
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.909876398649003e-05,
+      "loss": 0.0026,
+      "step": 16210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.908168147794285e-05,
+      "loss": 0.0038,
+      "step": 16211
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.906460015386554e-05,
+      "loss": 0.0006,
+      "step": 16212
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.904752001454328e-05,
+      "loss": 0.001,
+      "step": 16213
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.90304410602614e-05,
+      "loss": 0.0023,
+      "step": 16214
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.901336329130514e-05,
+      "loss": 0.0017,
+      "step": 16215
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.899628670795963e-05,
+      "loss": 0.0091,
+      "step": 16216
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.897921131051026e-05,
+      "loss": 0.0022,
+      "step": 16217
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.8962137099242e-05,
+      "loss": 0.0184,
+      "step": 16218
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.894506407444028e-05,
+      "loss": 0.006,
+      "step": 16219
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.892799223639003e-05,
+      "loss": 0.0067,
+      "step": 16220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.89109215853765e-05,
+      "loss": 0.0124,
+      "step": 16221
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.889385212168473e-05,
+      "loss": 0.0042,
+      "step": 16222
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.887678384559989e-05,
+      "loss": 0.0016,
+      "step": 16223
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.885971675740699e-05,
+      "loss": 0.0049,
+      "step": 16224
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.884265085739111e-05,
+      "loss": 0.0025,
+      "step": 16225
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.882558614583729e-05,
+      "loss": 0.0005,
+      "step": 16226
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.880852262303053e-05,
+      "loss": 0.0037,
+      "step": 16227
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.879146028925582e-05,
+      "loss": 0.0008,
+      "step": 16228
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.877439914479815e-05,
+      "loss": 0.0044,
+      "step": 16229
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.87573391899425e-05,
+      "loss": 0.0076,
+      "step": 16230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.874028042497366e-05,
+      "loss": 0.0046,
+      "step": 16231
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.872322285017676e-05,
+      "loss": 0.0022,
+      "step": 16232
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.870616646583648e-05,
+      "loss": 0.0149,
+      "step": 16233
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.86891112722379e-05,
+      "loss": 0.0032,
+      "step": 16234
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.867205726966573e-05,
+      "loss": 0.006,
+      "step": 16235
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.865500445840477e-05,
+      "loss": 0.0108,
+      "step": 16236
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.863795283874004e-05,
+      "loss": 0.0086,
+      "step": 16237
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.862090241095612e-05,
+      "loss": 0.0014,
+      "step": 16238
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.860385317533789e-05,
+      "loss": 0.0043,
+      "step": 16239
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.858680513217008e-05,
+      "loss": 0.0043,
+      "step": 16240
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.856975828173741e-05,
+      "loss": 0.0077,
+      "step": 16241
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.855271262432464e-05,
+      "loss": 0.0069,
+      "step": 16242
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.85356681602164e-05,
+      "loss": 0.0079,
+      "step": 16243
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.851862488969741e-05,
+      "loss": 0.0003,
+      "step": 16244
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.850158281305233e-05,
+      "loss": 0.0043,
+      "step": 16245
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.848454193056576e-05,
+      "loss": 0.0021,
+      "step": 16246
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.846750224252236e-05,
+      "loss": 0.0013,
+      "step": 16247
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.845046374920674e-05,
+      "loss": 0.0059,
+      "step": 16248
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.843342645090333e-05,
+      "loss": 0.0008,
+      "step": 16249
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.84163903478969e-05,
+      "loss": 0.0011,
+      "step": 16250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.839935544047174e-05,
+      "loss": 0.0003,
+      "step": 16251
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.838232172891265e-05,
+      "loss": 0.0198,
+      "step": 16252
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.836528921350391e-05,
+      "loss": 0.001,
+      "step": 16253
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.834825789453005e-05,
+      "loss": 0.0075,
+      "step": 16254
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.833122777227551e-05,
+      "loss": 0.0032,
+      "step": 16255
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.831419884702478e-05,
+      "loss": 0.0071,
+      "step": 16256
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.829717111906225e-05,
+      "loss": 0.0064,
+      "step": 16257
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.82801445886723e-05,
+      "loss": 0.0042,
+      "step": 16258
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.826311925613931e-05,
+      "loss": 0.0009,
+      "step": 16259
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.824609512174766e-05,
+      "loss": 0.0063,
+      "step": 16260
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.822907218578171e-05,
+      "loss": 0.0052,
+      "step": 16261
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.821205044852562e-05,
+      "loss": 0.009,
+      "step": 16262
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.81950299102639e-05,
+      "loss": 0.0055,
+      "step": 16263
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.817801057128065e-05,
+      "loss": 0.001,
+      "step": 16264
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.81609924318602e-05,
+      "loss": 0.002,
+      "step": 16265
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.814397549228688e-05,
+      "loss": 0.0051,
+      "step": 16266
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.812695975284469e-05,
+      "loss": 0.0025,
+      "step": 16267
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.810994521381803e-05,
+      "loss": 0.0084,
+      "step": 16268
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.809293187549089e-05,
+      "loss": 0.0023,
+      "step": 16269
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.807591973814764e-05,
+      "loss": 0.0028,
+      "step": 16270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.805890880207223e-05,
+      "loss": 0.0032,
+      "step": 16271
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.804189906754884e-05,
+      "loss": 0.0045,
+      "step": 16272
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.802489053486158e-05,
+      "loss": 0.0094,
+      "step": 16273
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.800788320429448e-05,
+      "loss": 0.0147,
+      "step": 16274
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.799087707613166e-05,
+      "loss": 0.0113,
+      "step": 16275
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.79738721506571e-05,
+      "loss": 0.0043,
+      "step": 16276
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.795686842815482e-05,
+      "loss": 0.0109,
+      "step": 16277
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.793986590890887e-05,
+      "loss": 0.0052,
+      "step": 16278
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.792286459320315e-05,
+      "loss": 0.0029,
+      "step": 16279
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.790586448132167e-05,
+      "loss": 0.0109,
+      "step": 16280
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.788886557354837e-05,
+      "loss": 0.0048,
+      "step": 16281
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.787186787016704e-05,
+      "loss": 0.0027,
+      "step": 16282
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.785487137146178e-05,
+      "loss": 0.0076,
+      "step": 16283
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.783787607771628e-05,
+      "loss": 0.0049,
+      "step": 16284
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.782088198921448e-05,
+      "loss": 0.0122,
+      "step": 16285
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.78038891062402e-05,
+      "loss": 0.0023,
+      "step": 16286
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.778689742907725e-05,
+      "loss": 0.0056,
+      "step": 16287
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.776990695800941e-05,
+      "loss": 0.001,
+      "step": 16288
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.775291769332047e-05,
+      "loss": 0.0032,
+      "step": 16289
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.77359296352942e-05,
+      "loss": 0.0012,
+      "step": 16290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.771894278421431e-05,
+      "loss": 0.0004,
+      "step": 16291
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.77019571403645e-05,
+      "loss": 0.0021,
+      "step": 16292
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.768497270402849e-05,
+      "loss": 0.0154,
+      "step": 16293
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.766798947548999e-05,
+      "loss": 0.0031,
+      "step": 16294
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.76510074550325e-05,
+      "loss": 0.0006,
+      "step": 16295
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.763402664293979e-05,
+      "loss": 0.0011,
+      "step": 16296
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.761704703949545e-05,
+      "loss": 0.0029,
+      "step": 16297
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.760006864498305e-05,
+      "loss": 0.002,
+      "step": 16298
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.758309145968624e-05,
+      "loss": 0.0048,
+      "step": 16299
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.756611548388837e-05,
+      "loss": 0.0104,
+      "step": 16300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.754914071787323e-05,
+      "loss": 0.0047,
+      "step": 16301
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.753216716192406e-05,
+      "loss": 0.0002,
+      "step": 16302
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.751519481632462e-05,
+      "loss": 0.0034,
+      "step": 16303
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.749822368135821e-05,
+      "loss": 0.0048,
+      "step": 16304
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.74812537573083e-05,
+      "loss": 0.0023,
+      "step": 16305
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.746428504445835e-05,
+      "loss": 0.0061,
+      "step": 16306
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.744731754309176e-05,
+      "loss": 0.0096,
+      "step": 16307
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.743035125349192e-05,
+      "loss": 0.0087,
+      "step": 16308
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.74133861759422e-05,
+      "loss": 0.0165,
+      "step": 16309
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.739642231072594e-05,
+      "loss": 0.0022,
+      "step": 16310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.737945965812648e-05,
+      "loss": 0.0004,
+      "step": 16311
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.736249821842719e-05,
+      "loss": 0.007,
+      "step": 16312
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.734553799191118e-05,
+      "loss": 0.0099,
+      "step": 16313
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.732857897886193e-05,
+      "loss": 0.0026,
+      "step": 16314
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.731162117956246e-05,
+      "loss": 0.0031,
+      "step": 16315
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.729466459429629e-05,
+      "loss": 0.0026,
+      "step": 16316
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.727770922334638e-05,
+      "loss": 0.0087,
+      "step": 16317
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.726075506699592e-05,
+      "loss": 0.0031,
+      "step": 16318
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.72438021255283e-05,
+      "loss": 0.0017,
+      "step": 16319
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.722685039922639e-05,
+      "loss": 0.0011,
+      "step": 16320
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.720989988837356e-05,
+      "loss": 0.002,
+      "step": 16321
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.719295059325274e-05,
+      "loss": 0.0034,
+      "step": 16322
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.71760025141471e-05,
+      "loss": 0.0027,
+      "step": 16323
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.715905565133968e-05,
+      "loss": 0.0019,
+      "step": 16324
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.714211000511353e-05,
+      "loss": 0.0043,
+      "step": 16325
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.712516557575164e-05,
+      "loss": 0.0126,
+      "step": 16326
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.710822236353707e-05,
+      "loss": 0.0005,
+      "step": 16327
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.709128036875279e-05,
+      "loss": 0.0002,
+      "step": 16328
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.707433959168173e-05,
+      "loss": 0.0049,
+      "step": 16329
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.705740003260687e-05,
+      "loss": 0.0061,
+      "step": 16330
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.704046169181112e-05,
+      "loss": 0.0003,
+      "step": 16331
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.702352456957743e-05,
+      "loss": 0.0042,
+      "step": 16332
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.700658866618851e-05,
+      "loss": 0.0033,
+      "step": 16333
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.698965398192748e-05,
+      "loss": 0.0008,
+      "step": 16334
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.697272051707697e-05,
+      "loss": 0.0056,
+      "step": 16335
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.695578827191988e-05,
+      "loss": 0.0106,
+      "step": 16336
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.693885724673902e-05,
+      "loss": 0.0004,
+      "step": 16337
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.692192744181716e-05,
+      "loss": 0.0021,
+      "step": 16338
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.690499885743704e-05,
+      "loss": 0.0009,
+      "step": 16339
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.688807149388143e-05,
+      "loss": 0.0021,
+      "step": 16340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.6871145351433e-05,
+      "loss": 0.0003,
+      "step": 16341
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.685422043037451e-05,
+      "loss": 0.0149,
+      "step": 16342
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.683729673098861e-05,
+      "loss": 0.0029,
+      "step": 16343
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.682037425355795e-05,
+      "loss": 0.0016,
+      "step": 16344
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.680345299836524e-05,
+      "loss": 0.0329,
+      "step": 16345
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.678653296569292e-05,
+      "loss": 0.0062,
+      "step": 16346
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.676961415582382e-05,
+      "loss": 0.0033,
+      "step": 16347
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.675269656904027e-05,
+      "loss": 0.0077,
+      "step": 16348
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.6735780205625e-05,
+      "loss": 0.0034,
+      "step": 16349
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.671886506586057e-05,
+      "loss": 0.001,
+      "step": 16350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.670195115002931e-05,
+      "loss": 0.0066,
+      "step": 16351
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.668503845841393e-05,
+      "loss": 0.0017,
+      "step": 16352
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.66681269912967e-05,
+      "loss": 0.0008,
+      "step": 16353
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.665121674896026e-05,
+      "loss": 0.0006,
+      "step": 16354
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.663430773168693e-05,
+      "loss": 0.0049,
+      "step": 16355
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.661739993975914e-05,
+      "loss": 0.0003,
+      "step": 16356
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.660049337345927e-05,
+      "loss": 0.0025,
+      "step": 16357
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.658358803306974e-05,
+      "loss": 0.0135,
+      "step": 16358
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.656668391887287e-05,
+      "loss": 0.0004,
+      "step": 16359
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.654978103115097e-05,
+      "loss": 0.0074,
+      "step": 16360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.653287937018637e-05,
+      "loss": 0.0112,
+      "step": 16361
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.651597893626138e-05,
+      "loss": 0.0082,
+      "step": 16362
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.649907972965833e-05,
+      "loss": 0.0032,
+      "step": 16363
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.648218175065925e-05,
+      "loss": 0.0019,
+      "step": 16364
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.646528499954663e-05,
+      "loss": 0.0056,
+      "step": 16365
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.644838947660244e-05,
+      "loss": 0.0012,
+      "step": 16366
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.64314951821091e-05,
+      "loss": 0.0009,
+      "step": 16367
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.64146021163486e-05,
+      "loss": 0.0002,
+      "step": 16368
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.639771027960317e-05,
+      "loss": 0.0058,
+      "step": 16369
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.63808196721549e-05,
+      "loss": 0.0016,
+      "step": 16370
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.636393029428583e-05,
+      "loss": 0.0145,
+      "step": 16371
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.634704214627825e-05,
+      "loss": 0.0017,
+      "step": 16372
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.633015522841405e-05,
+      "loss": 0.0015,
+      "step": 16373
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.631326954097531e-05,
+      "loss": 0.0068,
+      "step": 16374
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.629638508424406e-05,
+      "loss": 0.0092,
+      "step": 16375
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.62795018585023e-05,
+      "loss": 0.003,
+      "step": 16376
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.626261986403203e-05,
+      "loss": 0.004,
+      "step": 16377
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.62457391011152e-05,
+      "loss": 0.0048,
+      "step": 16378
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.622885957003372e-05,
+      "loss": 0.0072,
+      "step": 16379
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.621198127106956e-05,
+      "loss": 0.0027,
+      "step": 16380
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.61951042045046e-05,
+      "loss": 0.0233,
+      "step": 16381
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.617822837062071e-05,
+      "loss": 0.0048,
+      "step": 16382
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.616135376969982e-05,
+      "loss": 0.01,
+      "step": 16383
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.614448040202359e-05,
+      "loss": 0.0009,
+      "step": 16384
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.612760826787406e-05,
+      "loss": 0.0056,
+      "step": 16385
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.611073736753284e-05,
+      "loss": 0.002,
+      "step": 16386
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.609386770128179e-05,
+      "loss": 0.0136,
+      "step": 16387
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.607699926940268e-05,
+      "loss": 0.0005,
+      "step": 16388
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.606013207217718e-05,
+      "loss": 0.0018,
+      "step": 16389
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.604326610988709e-05,
+      "loss": 0.0197,
+      "step": 16390
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.602640138281404e-05,
+      "loss": 0.0004,
+      "step": 16391
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.600953789123971e-05,
+      "loss": 0.0056,
+      "step": 16392
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.599267563544577e-05,
+      "loss": 0.0033,
+      "step": 16393
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.597581461571385e-05,
+      "loss": 0.0044,
+      "step": 16394
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.595895483232556e-05,
+      "loss": 0.007,
+      "step": 16395
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.594209628556254e-05,
+      "loss": 0.0042,
+      "step": 16396
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.59252389757062e-05,
+      "loss": 0.0058,
+      "step": 16397
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.590838290303829e-05,
+      "loss": 0.0137,
+      "step": 16398
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.589152806784016e-05,
+      "loss": 0.0014,
+      "step": 16399
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.587467447039349e-05,
+      "loss": 0.008,
+      "step": 16400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.585782211097962e-05,
+      "loss": 0.001,
+      "step": 16401
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.584097098988003e-05,
+      "loss": 0.0065,
+      "step": 16402
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.582412110737632e-05,
+      "loss": 0.0027,
+      "step": 16403
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.580727246374968e-05,
+      "loss": 0.0028,
+      "step": 16404
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.579042505928176e-05,
+      "loss": 0.0056,
+      "step": 16405
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.577357889425376e-05,
+      "loss": 0.0072,
+      "step": 16406
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.575673396894711e-05,
+      "loss": 0.0074,
+      "step": 16407
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.573989028364317e-05,
+      "loss": 0.0095,
+      "step": 16408
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.572304783862319e-05,
+      "loss": 0.0054,
+      "step": 16409
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.570620663416857e-05,
+      "loss": 0.0032,
+      "step": 16410
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.568936667056049e-05,
+      "loss": 0.0097,
+      "step": 16411
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.567252794808029e-05,
+      "loss": 0.0018,
+      "step": 16412
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.565569046700918e-05,
+      "loss": 0.0006,
+      "step": 16413
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.563885422762841e-05,
+      "loss": 0.0056,
+      "step": 16414
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.562201923021902e-05,
+      "loss": 0.0221,
+      "step": 16415
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.560518547506245e-05,
+      "loss": 0.0027,
+      "step": 16416
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.558835296243959e-05,
+      "loss": 0.0003,
+      "step": 16417
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.55715216926318e-05,
+      "loss": 0.014,
+      "step": 16418
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.555469166592003e-05,
+      "loss": 0.0024,
+      "step": 16419
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.553786288258547e-05,
+      "loss": 0.0024,
+      "step": 16420
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.552103534290913e-05,
+      "loss": 0.0011,
+      "step": 16421
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.550420904717208e-05,
+      "loss": 0.0061,
+      "step": 16422
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.548738399565537e-05,
+      "loss": 0.0056,
+      "step": 16423
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.547056018864002e-05,
+      "loss": 0.0057,
+      "step": 16424
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.545373762640697e-05,
+      "loss": 0.0285,
+      "step": 16425
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.543691630923723e-05,
+      "loss": 0.0061,
+      "step": 16426
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.542009623741172e-05,
+      "loss": 0.0035,
+      "step": 16427
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.54032774112114e-05,
+      "loss": 0.0045,
+      "step": 16428
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.538645983091721e-05,
+      "loss": 0.0049,
+      "step": 16429
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.536964349680988e-05,
+      "loss": 0.0031,
+      "step": 16430
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.535282840917042e-05,
+      "loss": 0.0085,
+      "step": 16431
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.53360145682797e-05,
+      "loss": 0.0062,
+      "step": 16432
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.531920197441834e-05,
+      "loss": 0.0035,
+      "step": 16433
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.530239062786742e-05,
+      "loss": 0.0031,
+      "step": 16434
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.528558052890744e-05,
+      "loss": 0.0041,
+      "step": 16435
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.526877167781943e-05,
+      "loss": 0.0074,
+      "step": 16436
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.525196407488395e-05,
+      "loss": 0.0029,
+      "step": 16437
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.523515772038175e-05,
+      "loss": 0.0052,
+      "step": 16438
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.521835261459354e-05,
+      "loss": 0.0025,
+      "step": 16439
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.520154875780001e-05,
+      "loss": 0.0008,
+      "step": 16440
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.518474615028181e-05,
+      "loss": 0.0041,
+      "step": 16441
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.516794479231957e-05,
+      "loss": 0.0217,
+      "step": 16442
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.51511446841939e-05,
+      "loss": 0.0025,
+      "step": 16443
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.513434582618541e-05,
+      "loss": 0.0033,
+      "step": 16444
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.511754821857467e-05,
+      "loss": 0.001,
+      "step": 16445
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.510075186164221e-05,
+      "loss": 0.002,
+      "step": 16446
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.508395675566864e-05,
+      "loss": 0.0027,
+      "step": 16447
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.506716290093431e-05,
+      "loss": 0.0053,
+      "step": 16448
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.505037029771989e-05,
+      "loss": 0.0056,
+      "step": 16449
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.503357894630566e-05,
+      "loss": 0.0074,
+      "step": 16450
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.501678884697227e-05,
+      "loss": 0.006,
+      "step": 16451
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.500000000000002e-05,
+      "loss": 0.0017,
+      "step": 16452
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.498321240566933e-05,
+      "loss": 0.0015,
+      "step": 16453
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.496642606426059e-05,
+      "loss": 0.0026,
+      "step": 16454
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.494964097605412e-05,
+      "loss": 0.0014,
+      "step": 16455
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.493285714133043e-05,
+      "loss": 0.0032,
+      "step": 16456
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.491607456036968e-05,
+      "loss": 0.0148,
+      "step": 16457
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.489929323345218e-05,
+      "loss": 0.0009,
+      "step": 16458
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.488251316085828e-05,
+      "loss": 0.0038,
+      "step": 16459
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.486573434286823e-05,
+      "loss": 0.0026,
+      "step": 16460
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.484895677976216e-05,
+      "loss": 0.004,
+      "step": 16461
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.483218047182041e-05,
+      "loss": 0.0119,
+      "step": 16462
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.481540541932315e-05,
+      "loss": 0.0022,
+      "step": 16463
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.479863162255051e-05,
+      "loss": 0.0093,
+      "step": 16464
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.478185908178276e-05,
+      "loss": 0.0044,
+      "step": 16465
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.476508779729982e-05,
+      "loss": 0.0062,
+      "step": 16466
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.474831776938205e-05,
+      "loss": 0.014,
+      "step": 16467
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.473154899830931e-05,
+      "loss": 0.0074,
+      "step": 16468
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.471478148436189e-05,
+      "loss": 0.0017,
+      "step": 16469
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.469801522781967e-05,
+      "loss": 0.0066,
+      "step": 16470
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.468125022896274e-05,
+      "loss": 0.0043,
+      "step": 16471
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.46644864880711e-05,
+      "loss": 0.0016,
+      "step": 16472
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.464772400542476e-05,
+      "loss": 0.0029,
+      "step": 16473
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.463096278130365e-05,
+      "loss": 0.0016,
+      "step": 16474
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.461420281598773e-05,
+      "loss": 0.001,
+      "step": 16475
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.459744410975694e-05,
+      "loss": 0.0087,
+      "step": 16476
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.458068666289116e-05,
+      "loss": 0.0066,
+      "step": 16477
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.456393047567025e-05,
+      "loss": 0.0028,
+      "step": 16478
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.454717554837413e-05,
+      "loss": 0.0031,
+      "step": 16479
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.453042188128265e-05,
+      "loss": 0.0069,
+      "step": 16480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.451366947467547e-05,
+      "loss": 0.0089,
+      "step": 16481
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.449691832883261e-05,
+      "loss": 0.0089,
+      "step": 16482
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.448016844403368e-05,
+      "loss": 0.0008,
+      "step": 16483
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.446341982055844e-05,
+      "loss": 0.0014,
+      "step": 16484
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.44466724586868e-05,
+      "loss": 0.0035,
+      "step": 16485
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.44299263586982e-05,
+      "loss": 0.0009,
+      "step": 16486
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.441318152087261e-05,
+      "loss": 0.0047,
+      "step": 16487
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.439643794548952e-05,
+      "loss": 0.0116,
+      "step": 16488
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.437969563282863e-05,
+      "loss": 0.0039,
+      "step": 16489
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.436295458316955e-05,
+      "loss": 0.0047,
+      "step": 16490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.434621479679191e-05,
+      "loss": 0.0026,
+      "step": 16491
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.432947627397528e-05,
+      "loss": 0.0035,
+      "step": 16492
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.431273901499926e-05,
+      "loss": 0.0061,
+      "step": 16493
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.429600302014335e-05,
+      "loss": 0.001,
+      "step": 16494
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.42792682896871e-05,
+      "loss": 0.0079,
+      "step": 16495
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.426253482391e-05,
+      "loss": 0.0037,
+      "step": 16496
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.424580262309152e-05,
+      "loss": 0.0055,
+      "step": 16497
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.42290716875112e-05,
+      "loss": 0.0027,
+      "step": 16498
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.421234201744828e-05,
+      "loss": 0.0032,
+      "step": 16499
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.419561361318245e-05,
+      "loss": 0.006,
+      "step": 16500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.417888647499284e-05,
+      "loss": 0.0209,
+      "step": 16501
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.416216060315905e-05,
+      "loss": 0.0015,
+      "step": 16502
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.414543599796029e-05,
+      "loss": 0.0005,
+      "step": 16503
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.412871265967593e-05,
+      "loss": 0.006,
+      "step": 16504
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.411199058858529e-05,
+      "loss": 0.0103,
+      "step": 16505
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.409526978496764e-05,
+      "loss": 0.0015,
+      "step": 16506
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.407855024910228e-05,
+      "loss": 0.0023,
+      "step": 16507
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.406183198126843e-05,
+      "loss": 0.0101,
+      "step": 16508
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.404511498174534e-05,
+      "loss": 0.0024,
+      "step": 16509
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.402839925081221e-05,
+      "loss": 0.0048,
+      "step": 16510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.401168478874827e-05,
+      "loss": 0.0087,
+      "step": 16511
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.399497159583253e-05,
+      "loss": 0.004,
+      "step": 16512
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.397825967234433e-05,
+      "loss": 0.009,
+      "step": 16513
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.396154901856258e-05,
+      "loss": 0.0112,
+      "step": 16514
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.394483963476655e-05,
+      "loss": 0.0015,
+      "step": 16515
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.392813152123534e-05,
+      "loss": 0.0015,
+      "step": 16516
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.391142467824779e-05,
+      "loss": 0.0096,
+      "step": 16517
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.38947191060832e-05,
+      "loss": 0.002,
+      "step": 16518
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.387801480502034e-05,
+      "loss": 0.0092,
+      "step": 16519
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.386131177533846e-05,
+      "loss": 0.0002,
+      "step": 16520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.38446100173163e-05,
+      "loss": 0.0047,
+      "step": 16521
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.382790953123293e-05,
+      "loss": 0.0028,
+      "step": 16522
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.381121031736727e-05,
+      "loss": 0.006,
+      "step": 16523
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.379451237599819e-05,
+      "loss": 0.006,
+      "step": 16524
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.377781570740461e-05,
+      "loss": 0.0003,
+      "step": 16525
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.376112031186538e-05,
+      "loss": 0.0013,
+      "step": 16526
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.374442618965937e-05,
+      "loss": 0.0258,
+      "step": 16527
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.372773334106537e-05,
+      "loss": 0.0009,
+      "step": 16528
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.371104176636219e-05,
+      "loss": 0.0048,
+      "step": 16529
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.369435146582864e-05,
+      "loss": 0.0067,
+      "step": 16530
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.36776624397435e-05,
+      "loss": 0.003,
+      "step": 16531
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.366097468838536e-05,
+      "loss": 0.0043,
+      "step": 16532
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.364428821203314e-05,
+      "loss": 0.0018,
+      "step": 16533
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.36276030109654e-05,
+      "loss": 0.003,
+      "step": 16534
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.361091908546085e-05,
+      "loss": 0.0078,
+      "step": 16535
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.359423643579815e-05,
+      "loss": 0.0076,
+      "step": 16536
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.357755506225586e-05,
+      "loss": 0.0053,
+      "step": 16537
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.356087496511277e-05,
+      "loss": 0.0058,
+      "step": 16538
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.35441961446473e-05,
+      "loss": 0.0026,
+      "step": 16539
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.352751860113808e-05,
+      "loss": 0.0014,
+      "step": 16540
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.351084233486366e-05,
+      "loss": 0.0003,
+      "step": 16541
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.349416734610255e-05,
+      "loss": 0.0063,
+      "step": 16542
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.347749363513326e-05,
+      "loss": 0.0006,
+      "step": 16543
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.346082120223427e-05,
+      "loss": 0.002,
+      "step": 16544
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.344415004768403e-05,
+      "loss": 0.0024,
+      "step": 16545
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.342748017176102e-05,
+      "loss": 0.0067,
+      "step": 16546
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.341081157474362e-05,
+      "loss": 0.0062,
+      "step": 16547
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.339414425691025e-05,
+      "loss": 0.0006,
+      "step": 16548
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.337747821853932e-05,
+      "loss": 0.0007,
+      "step": 16549
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.336081345990903e-05,
+      "loss": 0.0013,
+      "step": 16550
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.334414998129792e-05,
+      "loss": 0.0036,
+      "step": 16551
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.332748778298411e-05,
+      "loss": 0.0029,
+      "step": 16552
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.331082686524609e-05,
+      "loss": 0.0076,
+      "step": 16553
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.329416722836197e-05,
+      "loss": 0.0074,
+      "step": 16554
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.327750887261003e-05,
+      "loss": 0.004,
+      "step": 16555
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.326085179826854e-05,
+      "loss": 0.0044,
+      "step": 16556
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.324419600561567e-05,
+      "loss": 0.0053,
+      "step": 16557
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.322754149492962e-05,
+      "loss": 0.0015,
+      "step": 16558
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.321088826648853e-05,
+      "loss": 0.0032,
+      "step": 16559
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.319423632057057e-05,
+      "loss": 0.0007,
+      "step": 16560
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.317758565745384e-05,
+      "loss": 0.0255,
+      "step": 16561
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.316093627741649e-05,
+      "loss": 0.0032,
+      "step": 16562
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.314428818073645e-05,
+      "loss": 0.0059,
+      "step": 16563
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.312764136769198e-05,
+      "loss": 0.0089,
+      "step": 16564
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.31109958385609e-05,
+      "loss": 0.0023,
+      "step": 16565
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.309435159362143e-05,
+      "loss": 0.0055,
+      "step": 16566
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.30777086331514e-05,
+      "loss": 0.0036,
+      "step": 16567
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.306106695742879e-05,
+      "loss": 0.0093,
+      "step": 16568
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.304442656673171e-05,
+      "loss": 0.0058,
+      "step": 16569
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.302778746133784e-05,
+      "loss": 0.0072,
+      "step": 16570
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.301114964152534e-05,
+      "loss": 0.0106,
+      "step": 16571
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.29945131075719e-05,
+      "loss": 0.0034,
+      "step": 16572
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.297787785975547e-05,
+      "loss": 0.0003,
+      "step": 16573
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.296124389835386e-05,
+      "loss": 0.0072,
+      "step": 16574
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.29446112236449e-05,
+      "loss": 0.0068,
+      "step": 16575
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.292797983590638e-05,
+      "loss": 0.0015,
+      "step": 16576
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.291134973541608e-05,
+      "loss": 0.0058,
+      "step": 16577
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.289472092245175e-05,
+      "loss": 0.0045,
+      "step": 16578
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.287809339729114e-05,
+      "loss": 0.0047,
+      "step": 16579
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.286146716021198e-05,
+      "loss": 0.0077,
+      "step": 16580
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.284484221149184e-05,
+      "loss": 0.0062,
+      "step": 16581
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.28282185514086e-05,
+      "loss": 0.0009,
+      "step": 16582
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.281159618023963e-05,
+      "loss": 0.0004,
+      "step": 16583
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.279497509826284e-05,
+      "loss": 0.0029,
+      "step": 16584
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.277835530575564e-05,
+      "loss": 0.0023,
+      "step": 16585
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.276173680299568e-05,
+      "loss": 0.0183,
+      "step": 16586
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.274511959026051e-05,
+      "loss": 0.0062,
+      "step": 16587
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.272850366782767e-05,
+      "loss": 0.0038,
+      "step": 16588
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.27118890359747e-05,
+      "loss": 0.0027,
+      "step": 16589
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.269527569497905e-05,
+      "loss": 0.0029,
+      "step": 16590
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.267866364511821e-05,
+      "loss": 0.0042,
+      "step": 16591
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.266205288666966e-05,
+      "loss": 0.0431,
+      "step": 16592
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.264544341991081e-05,
+      "loss": 0.0045,
+      "step": 16593
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.262883524511908e-05,
+      "loss": 0.0059,
+      "step": 16594
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.261222836257188e-05,
+      "loss": 0.0033,
+      "step": 16595
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.259562277254646e-05,
+      "loss": 0.0087,
+      "step": 16596
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.25790184753203e-05,
+      "loss": 0.0052,
+      "step": 16597
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.256241547117067e-05,
+      "loss": 0.0043,
+      "step": 16598
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.254581376037488e-05,
+      "loss": 0.004,
+      "step": 16599
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.252921334321025e-05,
+      "loss": 0.0024,
+      "step": 16600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.25126142199539e-05,
+      "loss": 0.0058,
+      "step": 16601
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.249601639088326e-05,
+      "loss": 0.0081,
+      "step": 16602
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.247941985627532e-05,
+      "loss": 0.0021,
+      "step": 16603
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.246282461640753e-05,
+      "loss": 0.0067,
+      "step": 16604
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.244623067155687e-05,
+      "loss": 0.0024,
+      "step": 16605
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.242963802200056e-05,
+      "loss": 0.0023,
+      "step": 16606
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.241304666801572e-05,
+      "loss": 0.0127,
+      "step": 16607
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.239645660987944e-05,
+      "loss": 0.0031,
+      "step": 16608
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.237986784786882e-05,
+      "loss": 0.0039,
+      "step": 16609
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.236328038226095e-05,
+      "loss": 0.0026,
+      "step": 16610
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.234669421333283e-05,
+      "loss": 0.0019,
+      "step": 16611
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.233010934136149e-05,
+      "loss": 0.0044,
+      "step": 16612
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.2313525766624e-05,
+      "loss": 0.0025,
+      "step": 16613
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.229694348939718e-05,
+      "loss": 0.002,
+      "step": 16614
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.228036250995816e-05,
+      "loss": 0.0029,
+      "step": 16615
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.226378282858369e-05,
+      "loss": 0.0063,
+      "step": 16616
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.22472044455509e-05,
+      "loss": 0.0083,
+      "step": 16617
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.223062736113653e-05,
+      "loss": 0.0084,
+      "step": 16618
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.221405157561746e-05,
+      "loss": 0.008,
+      "step": 16619
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.219747708927056e-05,
+      "loss": 0.0063,
+      "step": 16620
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.218090390237259e-05,
+      "loss": 0.0023,
+      "step": 16621
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.216433201520055e-05,
+      "loss": 0.0062,
+      "step": 16622
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.214776142803102e-05,
+      "loss": 0.0138,
+      "step": 16623
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.213119214114083e-05,
+      "loss": 0.0004,
+      "step": 16624
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.211462415480673e-05,
+      "loss": 0.0113,
+      "step": 16625
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.209805746930544e-05,
+      "loss": 0.0051,
+      "step": 16626
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.20814920849136e-05,
+      "loss": 0.0019,
+      "step": 16627
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.206492800190796e-05,
+      "loss": 0.0102,
+      "step": 16628
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.20483652205651e-05,
+      "loss": 0.0039,
+      "step": 16629
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.203180374116172e-05,
+      "loss": 0.0075,
+      "step": 16630
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.201524356397444e-05,
+      "loss": 0.0022,
+      "step": 16631
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.199868468927969e-05,
+      "loss": 0.0119,
+      "step": 16632
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.198212711735424e-05,
+      "loss": 0.0121,
+      "step": 16633
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.196557084847444e-05,
+      "loss": 0.0069,
+      "step": 16634
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.194901588291701e-05,
+      "loss": 0.0026,
+      "step": 16635
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.193246222095829e-05,
+      "loss": 0.0066,
+      "step": 16636
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.191590986287481e-05,
+      "loss": 0.0006,
+      "step": 16637
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.189935880894303e-05,
+      "loss": 0.0043,
+      "step": 16638
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.18828090594394e-05,
+      "loss": 0.0032,
+      "step": 16639
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.18662606146403e-05,
+      "loss": 0.004,
+      "step": 16640
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.184971347482215e-05,
+      "loss": 0.0014,
+      "step": 16641
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.183316764026128e-05,
+      "loss": 0.0064,
+      "step": 16642
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.181662311123408e-05,
+      "loss": 0.0013,
+      "step": 16643
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.180007988801685e-05,
+      "loss": 0.0056,
+      "step": 16644
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.178353797088591e-05,
+      "loss": 0.0043,
+      "step": 16645
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.176699736011757e-05,
+      "loss": 0.0036,
+      "step": 16646
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.175045805598794e-05,
+      "loss": 0.0003,
+      "step": 16647
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.173392005877347e-05,
+      "loss": 0.0026,
+      "step": 16648
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.171738336875018e-05,
+      "loss": 0.0015,
+      "step": 16649
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.17008479861944e-05,
+      "loss": 0.0065,
+      "step": 16650
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.168431391138232e-05,
+      "loss": 0.0024,
+      "step": 16651
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.166778114458992e-05,
+      "loss": 0.0032,
+      "step": 16652
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.165124968609356e-05,
+      "loss": 0.0023,
+      "step": 16653
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.163471953616914e-05,
+      "loss": 0.0089,
+      "step": 16654
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.161819069509286e-05,
+      "loss": 0.0023,
+      "step": 16655
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.160166316314073e-05,
+      "loss": 0.0027,
+      "step": 16656
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.15851369405888e-05,
+      "loss": 0.0091,
+      "step": 16657
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.156861202771311e-05,
+      "loss": 0.0136,
+      "step": 16658
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.155208842478968e-05,
+      "loss": 0.0094,
+      "step": 16659
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.153556613209444e-05,
+      "loss": 0.0086,
+      "step": 16660
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.151904514990334e-05,
+      "loss": 0.0111,
+      "step": 16661
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.150252547849232e-05,
+      "loss": 0.0122,
+      "step": 16662
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.148600711813734e-05,
+      "loss": 0.0102,
+      "step": 16663
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.146949006911427e-05,
+      "loss": 0.0167,
+      "step": 16664
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.145297433169885e-05,
+      "loss": 0.0068,
+      "step": 16665
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.143645990616714e-05,
+      "loss": 0.0021,
+      "step": 16666
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.141994679279473e-05,
+      "loss": 0.0016,
+      "step": 16667
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.140343499185766e-05,
+      "loss": 0.0009,
+      "step": 16668
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.138692450363154e-05,
+      "loss": 0.0104,
+      "step": 16669
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.137041532839215e-05,
+      "loss": 0.0017,
+      "step": 16670
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.135390746641526e-05,
+      "loss": 0.0046,
+      "step": 16671
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.133740091797658e-05,
+      "loss": 0.0026,
+      "step": 16672
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.132089568335178e-05,
+      "loss": 0.004,
+      "step": 16673
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.130439176281653e-05,
+      "loss": 0.0018,
+      "step": 16674
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.128788915664649e-05,
+      "loss": 0.0009,
+      "step": 16675
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.12713878651173e-05,
+      "loss": 0.002,
+      "step": 16676
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.125488788850452e-05,
+      "loss": 0.0018,
+      "step": 16677
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.123838922708378e-05,
+      "loss": 0.0036,
+      "step": 16678
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.122189188113065e-05,
+      "loss": 0.0017,
+      "step": 16679
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.120539585092052e-05,
+      "loss": 0.0056,
+      "step": 16680
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.118890113672908e-05,
+      "loss": 0.0033,
+      "step": 16681
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.11724077388318e-05,
+      "loss": 0.0027,
+      "step": 16682
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.115591565750401e-05,
+      "loss": 0.005,
+      "step": 16683
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.113942489302135e-05,
+      "loss": 0.014,
+      "step": 16684
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.112293544565905e-05,
+      "loss": 0.007,
+      "step": 16685
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.110644731569273e-05,
+      "loss": 0.0016,
+      "step": 16686
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.108996050339763e-05,
+      "loss": 0.004,
+      "step": 16687
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.10734750090491e-05,
+      "loss": 0.009,
+      "step": 16688
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.105699083292256e-05,
+      "loss": 0.0079,
+      "step": 16689
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.104050797529328e-05,
+      "loss": 0.0065,
+      "step": 16690
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.102402643643655e-05,
+      "loss": 0.0043,
+      "step": 16691
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.100754621662765e-05,
+      "loss": 0.0074,
+      "step": 16692
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.099106731614185e-05,
+      "loss": 0.001,
+      "step": 16693
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.097458973525438e-05,
+      "loss": 0.0071,
+      "step": 16694
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.095811347424041e-05,
+      "loss": 0.0076,
+      "step": 16695
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.094163853337515e-05,
+      "loss": 0.002,
+      "step": 16696
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.092516491293383e-05,
+      "loss": 0.012,
+      "step": 16697
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.09086926131914e-05,
+      "loss": 0.002,
+      "step": 16698
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.08922216344232e-05,
+      "loss": 0.0095,
+      "step": 16699
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.087575197690413e-05,
+      "loss": 0.0006,
+      "step": 16700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.085928364090947e-05,
+      "loss": 0.0018,
+      "step": 16701
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.08428166267141e-05,
+      "loss": 0.0153,
+      "step": 16702
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.082635093459306e-05,
+      "loss": 0.0002,
+      "step": 16703
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.080988656482152e-05,
+      "loss": 0.0097,
+      "step": 16704
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.07934235176743e-05,
+      "loss": 0.0156,
+      "step": 16705
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.077696179342642e-05,
+      "loss": 0.0027,
+      "step": 16706
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.076050139235281e-05,
+      "loss": 0.0047,
+      "step": 16707
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.074404231472842e-05,
+      "loss": 0.0026,
+      "step": 16708
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.072758456082812e-05,
+      "loss": 0.001,
+      "step": 16709
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.071112813092686e-05,
+      "loss": 0.0072,
+      "step": 16710
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.069467302529931e-05,
+      "loss": 0.0018,
+      "step": 16711
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.067821924422048e-05,
+      "loss": 0.0057,
+      "step": 16712
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.066176678796512e-05,
+      "loss": 0.0016,
+      "step": 16713
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.0645315656808e-05,
+      "loss": 0.0042,
+      "step": 16714
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.062886585102398e-05,
+      "loss": 0.0173,
+      "step": 16715
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.061241737088759e-05,
+      "loss": 0.0013,
+      "step": 16716
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.059597021667382e-05,
+      "loss": 0.003,
+      "step": 16717
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.057952438865711e-05,
+      "loss": 0.0073,
+      "step": 16718
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.056307988711236e-05,
+      "loss": 0.0055,
+      "step": 16719
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.054663671231408e-05,
+      "loss": 0.0036,
+      "step": 16720
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.053019486453695e-05,
+      "loss": 0.0023,
+      "step": 16721
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.051375434405556e-05,
+      "loss": 0.0021,
+      "step": 16722
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.049731515114453e-05,
+      "loss": 0.0014,
+      "step": 16723
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.048087728607841e-05,
+      "loss": 0.0007,
+      "step": 16724
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.046444074913173e-05,
+      "loss": 0.002,
+      "step": 16725
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.044800554057902e-05,
+      "loss": 0.0034,
+      "step": 16726
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.04315716606948e-05,
+      "loss": 0.0014,
+      "step": 16727
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.041513910975351e-05,
+      "loss": 0.0152,
+      "step": 16728
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.039870788802963e-05,
+      "loss": 0.0002,
+      "step": 16729
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.038227799579765e-05,
+      "loss": 0.0027,
+      "step": 16730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.036584943333181e-05,
+      "loss": 0.0029,
+      "step": 16731
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.03494222009067e-05,
+      "loss": 0.004,
+      "step": 16732
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.033299629879655e-05,
+      "loss": 0.0119,
+      "step": 16733
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.031657172727567e-05,
+      "loss": 0.0012,
+      "step": 16734
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.030014848661858e-05,
+      "loss": 0.0029,
+      "step": 16735
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.028372657709934e-05,
+      "loss": 0.0048,
+      "step": 16736
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.026730599899246e-05,
+      "loss": 0.0055,
+      "step": 16737
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.0250886752572e-05,
+      "loss": 0.0056,
+      "step": 16738
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.023446883811227e-05,
+      "loss": 0.0032,
+      "step": 16739
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.021805225588751e-05,
+      "loss": 0.0027,
+      "step": 16740
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.020163700617184e-05,
+      "loss": 0.0077,
+      "step": 16741
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.018522308923948e-05,
+      "loss": 0.0087,
+      "step": 16742
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.016881050536455e-05,
+      "loss": 0.0017,
+      "step": 16743
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.01523992548212e-05,
+      "loss": 0.0077,
+      "step": 16744
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.013598933788349e-05,
+      "loss": 0.0083,
+      "step": 16745
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.01195807548255e-05,
+      "loss": 0.0033,
+      "step": 16746
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.01031735059213e-05,
+      "loss": 0.0073,
+      "step": 16747
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.008676759144496e-05,
+      "loss": 0.001,
+      "step": 16748
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.007036301167036e-05,
+      "loss": 0.0057,
+      "step": 16749
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.005395976687169e-05,
+      "loss": 0.0017,
+      "step": 16750
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.003755785732267e-05,
+      "loss": 0.0007,
+      "step": 16751
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.00211572832975e-05,
+      "loss": 0.0071,
+      "step": 16752
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.000475804506991e-05,
+      "loss": 0.0008,
+      "step": 16753
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.998836014291387e-05,
+      "loss": 0.0012,
+      "step": 16754
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.997196357710325e-05,
+      "loss": 0.0041,
+      "step": 16755
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.99555683479119e-05,
+      "loss": 0.0061,
+      "step": 16756
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.993917445561365e-05,
+      "loss": 0.0057,
+      "step": 16757
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.992278190048231e-05,
+      "loss": 0.0066,
+      "step": 16758
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.990639068279166e-05,
+      "loss": 0.0034,
+      "step": 16759
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.989000080281549e-05,
+      "loss": 0.0093,
+      "step": 16760
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.987361226082758e-05,
+      "loss": 0.004,
+      "step": 16761
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.985722505710147e-05,
+      "loss": 0.004,
+      "step": 16762
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.984083919191111e-05,
+      "loss": 0.001,
+      "step": 16763
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.982445466552992e-05,
+      "loss": 0.0046,
+      "step": 16764
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.980807147823173e-05,
+      "loss": 0.0033,
+      "step": 16765
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.97916896302902e-05,
+      "loss": 0.0117,
+      "step": 16766
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.977530912197874e-05,
+      "loss": 0.0044,
+      "step": 16767
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.975892995357115e-05,
+      "loss": 0.0173,
+      "step": 16768
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.974255212534078e-05,
+      "loss": 0.0069,
+      "step": 16769
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.972617563756141e-05,
+      "loss": 0.008,
+      "step": 16770
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.970980049050637e-05,
+      "loss": 0.0033,
+      "step": 16771
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.969342668444923e-05,
+      "loss": 0.0008,
+      "step": 16772
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.967705421966346e-05,
+      "loss": 0.0004,
+      "step": 16773
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.966068309642249e-05,
+      "loss": 0.0033,
+      "step": 16774
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.964431331499977e-05,
+      "loss": 0.0072,
+      "step": 16775
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.962794487566869e-05,
+      "loss": 0.0011,
+      "step": 16776
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.961157777870266e-05,
+      "loss": 0.0058,
+      "step": 16777
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.959521202437501e-05,
+      "loss": 0.0061,
+      "step": 16778
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.957884761295915e-05,
+      "loss": 0.0037,
+      "step": 16779
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.956248454472824e-05,
+      "loss": 0.0053,
+      "step": 16780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.954612281995579e-05,
+      "loss": 0.0023,
+      "step": 16781
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.952976243891482e-05,
+      "loss": 0.0038,
+      "step": 16782
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.951340340187885e-05,
+      "loss": 0.0052,
+      "step": 16783
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.949704570912093e-05,
+      "loss": 0.0108,
+      "step": 16784
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.948068936091428e-05,
+      "loss": 0.0109,
+      "step": 16785
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.946433435753211e-05,
+      "loss": 0.0031,
+      "step": 16786
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.944798069924753e-05,
+      "loss": 0.0044,
+      "step": 16787
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.943162838633383e-05,
+      "loss": 0.0005,
+      "step": 16788
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.941527741906397e-05,
+      "loss": 0.0152,
+      "step": 16789
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.939892779771108e-05,
+      "loss": 0.0119,
+      "step": 16790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.938257952254826e-05,
+      "loss": 0.0029,
+      "step": 16791
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.936623259384852e-05,
+      "loss": 0.0063,
+      "step": 16792
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.934988701188493e-05,
+      "loss": 0.0099,
+      "step": 16793
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.933354277693042e-05,
+      "loss": 0.0089,
+      "step": 16794
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.931719988925805e-05,
+      "loss": 0.0028,
+      "step": 16795
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.930085834914074e-05,
+      "loss": 0.0037,
+      "step": 16796
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.92845181568514e-05,
+      "loss": 0.0015,
+      "step": 16797
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.9268179312663e-05,
+      "loss": 0.0027,
+      "step": 16798
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.925184181684843e-05,
+      "loss": 0.0041,
+      "step": 16799
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.923550566968043e-05,
+      "loss": 0.0015,
+      "step": 16800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.921917087143205e-05,
+      "loss": 0.0081,
+      "step": 16801
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.920283742237587e-05,
+      "loss": 0.0043,
+      "step": 16802
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.918650532278496e-05,
+      "loss": 0.0022,
+      "step": 16803
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.917017457293188e-05,
+      "loss": 0.0032,
+      "step": 16804
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.915384517308945e-05,
+      "loss": 0.0004,
+      "step": 16805
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.913751712353045e-05,
+      "loss": 0.0013,
+      "step": 16806
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.912119042452752e-05,
+      "loss": 0.0026,
+      "step": 16807
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.91048650763534e-05,
+      "loss": 0.0015,
+      "step": 16808
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.908854107928073e-05,
+      "loss": 0.0106,
+      "step": 16809
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.907221843358215e-05,
+      "loss": 0.0014,
+      "step": 16810
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.90558971395303e-05,
+      "loss": 0.0118,
+      "step": 16811
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.903957719739779e-05,
+      "loss": 0.0041,
+      "step": 16812
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.902325860745707e-05,
+      "loss": 0.0101,
+      "step": 16813
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.900694136998089e-05,
+      "loss": 0.0079,
+      "step": 16814
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.899062548524158e-05,
+      "loss": 0.0042,
+      "step": 16815
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.897431095351182e-05,
+      "loss": 0.0119,
+      "step": 16816
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.895799777506398e-05,
+      "loss": 0.0167,
+      "step": 16817
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.89416859501705e-05,
+      "loss": 0.0023,
+      "step": 16818
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.892537547910397e-05,
+      "loss": 0.0101,
+      "step": 16819
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.890906636213661e-05,
+      "loss": 0.0023,
+      "step": 16820
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.889275859954103e-05,
+      "loss": 0.0147,
+      "step": 16821
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.887645219158941e-05,
+      "loss": 0.0032,
+      "step": 16822
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.886014713855418e-05,
+      "loss": 0.0091,
+      "step": 16823
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.884384344070765e-05,
+      "loss": 0.0094,
+      "step": 16824
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.882754109832214e-05,
+      "loss": 0.0073,
+      "step": 16825
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.88112401116699e-05,
+      "loss": 0.0011,
+      "step": 16826
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.879494048102321e-05,
+      "loss": 0.0031,
+      "step": 16827
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.877864220665431e-05,
+      "loss": 0.0036,
+      "step": 16828
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.876234528883538e-05,
+      "loss": 0.0015,
+      "step": 16829
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.87460497278387e-05,
+      "loss": 0.005,
+      "step": 16830
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.872975552393626e-05,
+      "loss": 0.0044,
+      "step": 16831
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.871346267740041e-05,
+      "loss": 0.0035,
+      "step": 16832
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.869717118850308e-05,
+      "loss": 0.0112,
+      "step": 16833
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.868088105751655e-05,
+      "loss": 0.0052,
+      "step": 16834
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.866459228471276e-05,
+      "loss": 0.0046,
+      "step": 16835
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.864830487036381e-05,
+      "loss": 0.0014,
+      "step": 16836
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.863201881474175e-05,
+      "loss": 0.0039,
+      "step": 16837
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.861573411811854e-05,
+      "loss": 0.0016,
+      "step": 16838
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.85994507807662e-05,
+      "loss": 0.0013,
+      "step": 16839
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.85831688029567e-05,
+      "loss": 0.004,
+      "step": 16840
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.856688818496196e-05,
+      "loss": 0.0029,
+      "step": 16841
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.855060892705391e-05,
+      "loss": 0.0069,
+      "step": 16842
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.853433102950443e-05,
+      "loss": 0.0118,
+      "step": 16843
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.85180544925854e-05,
+      "loss": 0.0168,
+      "step": 16844
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.850177931656872e-05,
+      "loss": 0.0087,
+      "step": 16845
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.848550550172604e-05,
+      "loss": 0.0014,
+      "step": 16846
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.846923304832936e-05,
+      "loss": 0.0043,
+      "step": 16847
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.845296195665038e-05,
+      "loss": 0.0027,
+      "step": 16848
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.843669222696087e-05,
+      "loss": 0.0048,
+      "step": 16849
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.842042385953262e-05,
+      "loss": 0.0041,
+      "step": 16850
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.840415685463718e-05,
+      "loss": 0.0038,
+      "step": 16851
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.838789121254645e-05,
+      "loss": 0.0079,
+      "step": 16852
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.837162693353193e-05,
+      "loss": 0.0039,
+      "step": 16853
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.835536401786532e-05,
+      "loss": 0.0135,
+      "step": 16854
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.833910246581827e-05,
+      "loss": 0.0048,
+      "step": 16855
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.832284227766233e-05,
+      "loss": 0.0038,
+      "step": 16856
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.83065834536691e-05,
+      "loss": 0.0022,
+      "step": 16857
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.829032599411016e-05,
+      "loss": 0.0009,
+      "step": 16858
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.827406989925697e-05,
+      "loss": 0.007,
+      "step": 16859
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.825781516938111e-05,
+      "loss": 0.0134,
+      "step": 16860
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.824156180475404e-05,
+      "loss": 0.0015,
+      "step": 16861
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.82253098056472e-05,
+      "loss": 0.012,
+      "step": 16862
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.820905917233212e-05,
+      "loss": 0.0013,
+      "step": 16863
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.819280990508003e-05,
+      "loss": 0.0069,
+      "step": 16864
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.817656200416255e-05,
+      "loss": 0.0007,
+      "step": 16865
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.81603154698508e-05,
+      "loss": 0.001,
+      "step": 16866
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.814407030241641e-05,
+      "loss": 0.0036,
+      "step": 16867
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.812782650213048e-05,
+      "loss": 0.0173,
+      "step": 16868
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.811158406926441e-05,
+      "loss": 0.0019,
+      "step": 16869
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.809534300408945e-05,
+      "loss": 0.0052,
+      "step": 16870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.807910330687682e-05,
+      "loss": 0.0024,
+      "step": 16871
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.806286497789791e-05,
+      "loss": 0.0057,
+      "step": 16872
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.804662801742377e-05,
+      "loss": 0.0038,
+      "step": 16873
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.803039242572563e-05,
+      "loss": 0.0027,
+      "step": 16874
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.801415820307467e-05,
+      "loss": 0.0048,
+      "step": 16875
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.799792534974203e-05,
+      "loss": 0.0015,
+      "step": 16876
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.798169386599881e-05,
+      "loss": 0.004,
+      "step": 16877
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.796546375211614e-05,
+      "loss": 0.0056,
+      "step": 16878
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.794923500836509e-05,
+      "loss": 0.0081,
+      "step": 16879
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.793300763501666e-05,
+      "loss": 0.0038,
+      "step": 16880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.7916781632342e-05,
+      "loss": 0.0037,
+      "step": 16881
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.79005570006119e-05,
+      "loss": 0.0052,
+      "step": 16882
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.78843337400976e-05,
+      "loss": 0.0264,
+      "step": 16883
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.786811185106981e-05,
+      "loss": 0.0091,
+      "step": 16884
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.785189133379971e-05,
+      "loss": 0.0135,
+      "step": 16885
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.783567218855802e-05,
+      "loss": 0.009,
+      "step": 16886
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.78194544156157e-05,
+      "loss": 0.0057,
+      "step": 16887
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.780323801524362e-05,
+      "loss": 0.0033,
+      "step": 16888
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.778702298771264e-05,
+      "loss": 0.0031,
+      "step": 16889
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.777080933329354e-05,
+      "loss": 0.0017,
+      "step": 16890
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.775459705225715e-05,
+      "loss": 0.0004,
+      "step": 16891
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.773838614487424e-05,
+      "loss": 0.0073,
+      "step": 16892
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.772217661141557e-05,
+      "loss": 0.0004,
+      "step": 16893
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.770596845215185e-05,
+      "loss": 0.0072,
+      "step": 16894
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.76897616673538e-05,
+      "loss": 0.0078,
+      "step": 16895
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.767355625729215e-05,
+      "loss": 0.0026,
+      "step": 16896
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.765735222223741e-05,
+      "loss": 0.001,
+      "step": 16897
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.764114956246045e-05,
+      "loss": 0.0044,
+      "step": 16898
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.762494827823162e-05,
+      "loss": 0.0081,
+      "step": 16899
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.760874836982171e-05,
+      "loss": 0.004,
+      "step": 16900
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.759254983750127e-05,
+      "loss": 0.0126,
+      "step": 16901
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.75763526815407e-05,
+      "loss": 0.0039,
+      "step": 16902
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.756015690221076e-05,
+      "loss": 0.0063,
+      "step": 16903
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.754396249978173e-05,
+      "loss": 0.0012,
+      "step": 16904
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.752776947452418e-05,
+      "loss": 0.0115,
+      "step": 16905
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.751157782670856e-05,
+      "loss": 0.0049,
+      "step": 16906
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.749538755660531e-05,
+      "loss": 0.0085,
+      "step": 16907
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.747919866448481e-05,
+      "loss": 0.0046,
+      "step": 16908
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.746301115061747e-05,
+      "loss": 0.0022,
+      "step": 16909
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.744682501527363e-05,
+      "loss": 0.0066,
+      "step": 16910
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.743064025872367e-05,
+      "loss": 0.0086,
+      "step": 16911
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.741445688123785e-05,
+      "loss": 0.0009,
+      "step": 16912
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.739827488308652e-05,
+      "loss": 0.0093,
+      "step": 16913
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.738209426453994e-05,
+      "loss": 0.001,
+      "step": 16914
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.736591502586824e-05,
+      "loss": 0.0059,
+      "step": 16915
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.734973716734187e-05,
+      "loss": 0.0187,
+      "step": 16916
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.733356068923079e-05,
+      "loss": 0.0064,
+      "step": 16917
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.731738559180539e-05,
+      "loss": 0.0037,
+      "step": 16918
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.730121187533566e-05,
+      "loss": 0.0125,
+      "step": 16919
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.728503954009181e-05,
+      "loss": 0.0014,
+      "step": 16920
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.726886858634393e-05,
+      "loss": 0.0008,
+      "step": 16921
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.725269901436213e-05,
+      "loss": 0.0094,
+      "step": 16922
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.723653082441642e-05,
+      "loss": 0.0007,
+      "step": 16923
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.722036401677691e-05,
+      "loss": 0.0128,
+      "step": 16924
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.720419859171355e-05,
+      "loss": 0.0016,
+      "step": 16925
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.718803454949637e-05,
+      "loss": 0.006,
+      "step": 16926
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.717187189039538e-05,
+      "loss": 0.0029,
+      "step": 16927
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.715571061468037e-05,
+      "loss": 0.0047,
+      "step": 16928
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.71395507226215e-05,
+      "loss": 0.0018,
+      "step": 16929
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.712339221448841e-05,
+      "loss": 0.0041,
+      "step": 16930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.710723509055119e-05,
+      "loss": 0.0018,
+      "step": 16931
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.709107935107965e-05,
+      "loss": 0.0009,
+      "step": 16932
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.707492499634348e-05,
+      "loss": 0.0035,
+      "step": 16933
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.705877202661272e-05,
+      "loss": 0.0026,
+      "step": 16934
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.70426204421569e-05,
+      "loss": 0.0072,
+      "step": 16935
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.702647024324604e-05,
+      "loss": 0.0034,
+      "step": 16936
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.701032143014967e-05,
+      "loss": 0.0031,
+      "step": 16937
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.69941740031376e-05,
+      "loss": 0.0108,
+      "step": 16938
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.697802796247951e-05,
+      "loss": 0.0027,
+      "step": 16939
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.696188330844508e-05,
+      "loss": 0.0008,
+      "step": 16940
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.694574004130393e-05,
+      "loss": 0.0063,
+      "step": 16941
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.69295981613257e-05,
+      "loss": 0.0106,
+      "step": 16942
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.691345766877999e-05,
+      "loss": 0.0051,
+      "step": 16943
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.689731856393636e-05,
+      "loss": 0.0019,
+      "step": 16944
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.68811808470644e-05,
+      "loss": 0.0057,
+      "step": 16945
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.68650445184336e-05,
+      "loss": 0.0018,
+      "step": 16946
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.684890957831354e-05,
+      "loss": 0.0047,
+      "step": 16947
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.683277602697356e-05,
+      "loss": 0.0014,
+      "step": 16948
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.681664386468331e-05,
+      "loss": 0.0032,
+      "step": 16949
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.6800513091712e-05,
+      "loss": 0.0056,
+      "step": 16950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.67843837083293e-05,
+      "loss": 0.0067,
+      "step": 16951
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.67682557148044e-05,
+      "loss": 0.0019,
+      "step": 16952
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.675212911140669e-05,
+      "loss": 0.0097,
+      "step": 16953
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.673600389840569e-05,
+      "loss": 0.0094,
+      "step": 16954
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.671988007607054e-05,
+      "loss": 0.0061,
+      "step": 16955
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.670375764467058e-05,
+      "loss": 0.0024,
+      "step": 16956
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.668763660447513e-05,
+      "loss": 0.0023,
+      "step": 16957
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.667151695575338e-05,
+      "loss": 0.0005,
+      "step": 16958
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.665539869877459e-05,
+      "loss": 0.0048,
+      "step": 16959
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.663928183380798e-05,
+      "loss": 0.0053,
+      "step": 16960
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.662316636112272e-05,
+      "loss": 0.0035,
+      "step": 16961
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.660705228098798e-05,
+      "loss": 0.0054,
+      "step": 16962
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.659093959367286e-05,
+      "loss": 0.0022,
+      "step": 16963
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.657482829944648e-05,
+      "loss": 0.0063,
+      "step": 16964
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.655871839857802e-05,
+      "loss": 0.0152,
+      "step": 16965
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.654260989133636e-05,
+      "loss": 0.0059,
+      "step": 16966
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.652650277799076e-05,
+      "loss": 0.0003,
+      "step": 16967
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.651039705881001e-05,
+      "loss": 0.0044,
+      "step": 16968
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.649429273406335e-05,
+      "loss": 0.0011,
+      "step": 16969
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.647818980401957e-05,
+      "loss": 0.0127,
+      "step": 16970
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.646208826894768e-05,
+      "loss": 0.0076,
+      "step": 16971
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.644598812911662e-05,
+      "loss": 0.006,
+      "step": 16972
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.642988938479526e-05,
+      "loss": 0.0061,
+      "step": 16973
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.641379203625249e-05,
+      "loss": 0.0041,
+      "step": 16974
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.639769608375718e-05,
+      "loss": 0.0029,
+      "step": 16975
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.638160152757817e-05,
+      "loss": 0.0014,
+      "step": 16976
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.636550836798424e-05,
+      "loss": 0.0054,
+      "step": 16977
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.634941660524424e-05,
+      "loss": 0.0017,
+      "step": 16978
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.633332623962679e-05,
+      "loss": 0.0054,
+      "step": 16979
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.631723727140082e-05,
+      "loss": 0.0055,
+      "step": 16980
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.630114970083486e-05,
+      "loss": 0.012,
+      "step": 16981
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.628506352819779e-05,
+      "loss": 0.009,
+      "step": 16982
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.626897875375813e-05,
+      "loss": 0.0045,
+      "step": 16983
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.625289537778452e-05,
+      "loss": 0.0004,
+      "step": 16984
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.623681340054575e-05,
+      "loss": 0.0046,
+      "step": 16985
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.62207328223102e-05,
+      "loss": 0.0051,
+      "step": 16986
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.620465364334667e-05,
+      "loss": 0.0006,
+      "step": 16987
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.618857586392355e-05,
+      "loss": 0.0069,
+      "step": 16988
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.617249948430941e-05,
+      "loss": 0.0025,
+      "step": 16989
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.615642450477276e-05,
+      "loss": 0.0061,
+      "step": 16990
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.614035092558209e-05,
+      "loss": 0.0016,
+      "step": 16991
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.612427874700586e-05,
+      "loss": 0.0007,
+      "step": 16992
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.61082079693125e-05,
+      "loss": 0.0235,
+      "step": 16993
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.609213859277043e-05,
+      "loss": 0.0019,
+      "step": 16994
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.607607061764804e-05,
+      "loss": 0.004,
+      "step": 16995
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.606000404421365e-05,
+      "loss": 0.0043,
+      "step": 16996
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.604393887273568e-05,
+      "loss": 0.005,
+      "step": 16997
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.602787510348245e-05,
+      "loss": 0.0082,
+      "step": 16998
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.60118127367221e-05,
+      "loss": 0.0076,
+      "step": 16999
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.599575177272313e-05,
+      "loss": 0.0004,
+      "step": 17000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.597969221175363e-05,
+      "loss": 0.0113,
+      "step": 17001
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.596363405408185e-05,
+      "loss": 0.002,
+      "step": 17002
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.594757729997602e-05,
+      "loss": 0.0012,
+      "step": 17003
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.593152194970431e-05,
+      "loss": 0.0014,
+      "step": 17004
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.591546800353489e-05,
+      "loss": 0.0055,
+      "step": 17005
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.589941546173586e-05,
+      "loss": 0.0041,
+      "step": 17006
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.588336432457533e-05,
+      "loss": 0.0031,
+      "step": 17007
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.58673145923214e-05,
+      "loss": 0.0025,
+      "step": 17008
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.585126626524214e-05,
+      "loss": 0.0045,
+      "step": 17009
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.583521934360558e-05,
+      "loss": 0.0028,
+      "step": 17010
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.581917382767978e-05,
+      "loss": 0.007,
+      "step": 17011
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.580312971773256e-05,
+      "loss": 0.0027,
+      "step": 17012
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.578708701403207e-05,
+      "loss": 0.006,
+      "step": 17013
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.577104571684619e-05,
+      "loss": 0.013,
+      "step": 17014
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.575500582644284e-05,
+      "loss": 0.0014,
+      "step": 17015
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.573896734308996e-05,
+      "loss": 0.0024,
+      "step": 17016
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.572293026705529e-05,
+      "loss": 0.003,
+      "step": 17017
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.570689459860687e-05,
+      "loss": 0.003,
+      "step": 17018
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.56908603380123e-05,
+      "loss": 0.0017,
+      "step": 17019
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.567482748553964e-05,
+      "loss": 0.0013,
+      "step": 17020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.565879604145647e-05,
+      "loss": 0.0078,
+      "step": 17021
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.564276600603061e-05,
+      "loss": 0.0027,
+      "step": 17022
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.56267373795298e-05,
+      "loss": 0.0036,
+      "step": 17023
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.561071016222174e-05,
+      "loss": 0.0023,
+      "step": 17024
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.559468435437411e-05,
+      "loss": 0.0004,
+      "step": 17025
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.557865995625458e-05,
+      "loss": 0.0056,
+      "step": 17026
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.55626369681308e-05,
+      "loss": 0.0008,
+      "step": 17027
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.554661539027033e-05,
+      "loss": 0.0079,
+      "step": 17028
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.553059522294089e-05,
+      "loss": 0.0045,
+      "step": 17029
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.551457646640984e-05,
+      "loss": 0.0047,
+      "step": 17030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.549855912094494e-05,
+      "loss": 0.0055,
+      "step": 17031
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.548254318681352e-05,
+      "loss": 0.0031,
+      "step": 17032
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.546652866428325e-05,
+      "loss": 0.0058,
+      "step": 17033
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.545051555362149e-05,
+      "loss": 0.0036,
+      "step": 17034
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.54345038550957e-05,
+      "loss": 0.0035,
+      "step": 17035
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.541849356897335e-05,
+      "loss": 0.0056,
+      "step": 17036
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.540248469552174e-05,
+      "loss": 0.0041,
+      "step": 17037
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.538647723500844e-05,
+      "loss": 0.004,
+      "step": 17038
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.537047118770063e-05,
+      "loss": 0.0014,
+      "step": 17039
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.53544665538657e-05,
+      "loss": 0.0036,
+      "step": 17040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.533846333377096e-05,
+      "loss": 0.0056,
+      "step": 17041
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.53224615276837e-05,
+      "loss": 0.0079,
+      "step": 17042
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.530646113587117e-05,
+      "loss": 0.0145,
+      "step": 17043
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.529046215860061e-05,
+      "loss": 0.0104,
+      "step": 17044
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.527446459613921e-05,
+      "loss": 0.0006,
+      "step": 17045
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.525846844875421e-05,
+      "loss": 0.0096,
+      "step": 17046
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.524247371671274e-05,
+      "loss": 0.0019,
+      "step": 17047
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.522648040028195e-05,
+      "loss": 0.0019,
+      "step": 17048
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.521048849972902e-05,
+      "loss": 0.0041,
+      "step": 17049
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.519449801532086e-05,
+      "loss": 0.0149,
+      "step": 17050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.517850894732479e-05,
+      "loss": 0.0036,
+      "step": 17051
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.516252129600767e-05,
+      "loss": 0.0034,
+      "step": 17052
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.514653506163659e-05,
+      "loss": 0.0028,
+      "step": 17053
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.513055024447853e-05,
+      "loss": 0.0017,
+      "step": 17054
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.51145668448005e-05,
+      "loss": 0.0029,
+      "step": 17055
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.509858486286942e-05,
+      "loss": 0.0037,
+      "step": 17056
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.508260429895227e-05,
+      "loss": 0.0031,
+      "step": 17057
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.506662515331589e-05,
+      "loss": 0.004,
+      "step": 17058
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.505064742622721e-05,
+      "loss": 0.0009,
+      "step": 17059
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.503467111795304e-05,
+      "loss": 0.0017,
+      "step": 17060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.501869622876028e-05,
+      "loss": 0.0021,
+      "step": 17061
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.500272275891574e-05,
+      "loss": 0.0014,
+      "step": 17062
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.498675070868608e-05,
+      "loss": 0.0047,
+      "step": 17063
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.497078007833825e-05,
+      "loss": 0.0033,
+      "step": 17064
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.495481086813882e-05,
+      "loss": 0.0028,
+      "step": 17065
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.49388430783546e-05,
+      "loss": 0.002,
+      "step": 17066
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.492287670925234e-05,
+      "loss": 0.0063,
+      "step": 17067
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.490691176109852e-05,
+      "loss": 0.0056,
+      "step": 17068
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.489094823416001e-05,
+      "loss": 0.0073,
+      "step": 17069
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.48749861287032e-05,
+      "loss": 0.0119,
+      "step": 17070
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.485902544499493e-05,
+      "loss": 0.0002,
+      "step": 17071
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.484306618330159e-05,
+      "loss": 0.0116,
+      "step": 17072
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.482710834388979e-05,
+      "loss": 0.0008,
+      "step": 17073
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.481115192702605e-05,
+      "loss": 0.0011,
+      "step": 17074
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.479519693297688e-05,
+      "loss": 0.0055,
+      "step": 17075
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.477924336200877e-05,
+      "loss": 0.0059,
+      "step": 17076
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.476329121438812e-05,
+      "loss": 0.003,
+      "step": 17077
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.474734049038145e-05,
+      "loss": 0.0001,
+      "step": 17078
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.47313911902551e-05,
+      "loss": 0.0082,
+      "step": 17079
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.471544331427553e-05,
+      "loss": 0.0053,
+      "step": 17080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.469949686270893e-05,
+      "loss": 0.0013,
+      "step": 17081
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.468355183582186e-05,
+      "loss": 0.0004,
+      "step": 17082
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.466760823388041e-05,
+      "loss": 0.0086,
+      "step": 17083
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.465166605715108e-05,
+      "loss": 0.009,
+      "step": 17084
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.463572530590001e-05,
+      "loss": 0.004,
+      "step": 17085
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.461978598039344e-05,
+      "loss": 0.0021,
+      "step": 17086
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.46038480808976e-05,
+      "loss": 0.0026,
+      "step": 17087
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.458791160767872e-05,
+      "loss": 0.0072,
+      "step": 17088
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.457197656100294e-05,
+      "loss": 0.0065,
+      "step": 17089
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.455604294113639e-05,
+      "loss": 0.0013,
+      "step": 17090
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.45401107483452e-05,
+      "loss": 0.0008,
+      "step": 17091
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.45241799828955e-05,
+      "loss": 0.0079,
+      "step": 17092
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.450825064505334e-05,
+      "loss": 0.0111,
+      "step": 17093
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.449232273508477e-05,
+      "loss": 0.0086,
+      "step": 17094
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.447639625325587e-05,
+      "loss": 0.0018,
+      "step": 17095
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.446047119983248e-05,
+      "loss": 0.0041,
+      "step": 17096
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.444454757508075e-05,
+      "loss": 0.016,
+      "step": 17097
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.442862537926655e-05,
+      "loss": 0.0006,
+      "step": 17098
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.441270461265587e-05,
+      "loss": 0.0047,
+      "step": 17099
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.43967852755146e-05,
+      "loss": 0.0103,
+      "step": 17100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.438086736810852e-05,
+      "loss": 0.0039,
+      "step": 17101
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.436495089070367e-05,
+      "loss": 0.0072,
+      "step": 17102
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.434903584356572e-05,
+      "loss": 0.0056,
+      "step": 17103
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.433312222696058e-05,
+      "loss": 0.0047,
+      "step": 17104
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.431721004115397e-05,
+      "loss": 0.0037,
+      "step": 17105
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.430129928641171e-05,
+      "loss": 0.0024,
+      "step": 17106
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.428538996299954e-05,
+      "loss": 0.0127,
+      "step": 17107
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.426948207118312e-05,
+      "loss": 0.001,
+      "step": 17108
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.425357561122819e-05,
+      "loss": 0.0082,
+      "step": 17109
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.42376705834004e-05,
+      "loss": 0.0023,
+      "step": 17110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.422176698796542e-05,
+      "loss": 0.0022,
+      "step": 17111
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.420586482518882e-05,
+      "loss": 0.0068,
+      "step": 17112
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.418996409533631e-05,
+      "loss": 0.0019,
+      "step": 17113
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.417406479867326e-05,
+      "loss": 0.0023,
+      "step": 17114
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.415816693546543e-05,
+      "loss": 0.0067,
+      "step": 17115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.414227050597816e-05,
+      "loss": 0.0105,
+      "step": 17116
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.412637551047713e-05,
+      "loss": 0.0105,
+      "step": 17117
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.411048194922767e-05,
+      "loss": 0.0066,
+      "step": 17118
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.409458982249525e-05,
+      "loss": 0.0001,
+      "step": 17119
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.407869913054544e-05,
+      "loss": 0.0048,
+      "step": 17120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.406280987364345e-05,
+      "loss": 0.0073,
+      "step": 17121
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.404692205205486e-05,
+      "loss": 0.0007,
+      "step": 17122
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.403103566604485e-05,
+      "loss": 0.0031,
+      "step": 17123
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.401515071587881e-05,
+      "loss": 0.0064,
+      "step": 17124
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.399926720182209e-05,
+      "loss": 0.0059,
+      "step": 17125
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.398338512413992e-05,
+      "loss": 0.0006,
+      "step": 17126
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.396750448309761e-05,
+      "loss": 0.0032,
+      "step": 17127
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.395162527896036e-05,
+      "loss": 0.0121,
+      "step": 17128
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.393574751199339e-05,
+      "loss": 0.0062,
+      "step": 17129
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.391987118246191e-05,
+      "loss": 0.0003,
+      "step": 17130
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.390399629063112e-05,
+      "loss": 0.0026,
+      "step": 17131
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.388812283676598e-05,
+      "loss": 0.0019,
+      "step": 17132
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.387225082113186e-05,
+      "loss": 0.0062,
+      "step": 17133
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.385638024399362e-05,
+      "loss": 0.0035,
+      "step": 17134
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.384051110561655e-05,
+      "loss": 0.0039,
+      "step": 17135
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.382464340626554e-05,
+      "loss": 0.0046,
+      "step": 17136
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.380877714620563e-05,
+      "loss": 0.0006,
+      "step": 17137
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.379291232570184e-05,
+      "loss": 0.0046,
+      "step": 17138
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.377704894501915e-05,
+      "loss": 0.0013,
+      "step": 17139
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.37611870044225e-05,
+      "loss": 0.0022,
+      "step": 17140
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.374532650417681e-05,
+      "loss": 0.0004,
+      "step": 17141
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.3729467444547e-05,
+      "loss": 0.0032,
+      "step": 17142
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.371360982579792e-05,
+      "loss": 0.0095,
+      "step": 17143
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.369775364819443e-05,
+      "loss": 0.0008,
+      "step": 17144
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.368189891200137e-05,
+      "loss": 0.0042,
+      "step": 17145
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.366604561748357e-05,
+      "loss": 0.0021,
+      "step": 17146
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.365019376490568e-05,
+      "loss": 0.0021,
+      "step": 17147
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.363434335453267e-05,
+      "loss": 0.0012,
+      "step": 17148
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.361849438662907e-05,
+      "loss": 0.0063,
+      "step": 17149
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.360264686145964e-05,
+      "loss": 0.0051,
+      "step": 17150
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.358680077928919e-05,
+      "loss": 0.0073,
+      "step": 17151
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.357095614038219e-05,
+      "loss": 0.0041,
+      "step": 17152
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.355511294500346e-05,
+      "loss": 0.0091,
+      "step": 17153
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.353927119341747e-05,
+      "loss": 0.0125,
+      "step": 17154
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.352343088588887e-05,
+      "loss": 0.0026,
+      "step": 17155
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.350759202268221e-05,
+      "loss": 0.0009,
+      "step": 17156
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.349175460406201e-05,
+      "loss": 0.0086,
+      "step": 17157
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.347591863029282e-05,
+      "loss": 0.0007,
+      "step": 17158
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.34600841016391e-05,
+      "loss": 0.0051,
+      "step": 17159
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.344425101836534e-05,
+      "loss": 0.0011,
+      "step": 17160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.342841938073598e-05,
+      "loss": 0.0006,
+      "step": 17161
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.341258918901542e-05,
+      "loss": 0.0067,
+      "step": 17162
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.339676044346807e-05,
+      "loss": 0.0045,
+      "step": 17163
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.338093314435834e-05,
+      "loss": 0.0036,
+      "step": 17164
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.336510729195046e-05,
+      "loss": 0.0066,
+      "step": 17165
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.334928288650889e-05,
+      "loss": 0.0071,
+      "step": 17166
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.333345992829777e-05,
+      "loss": 0.0042,
+      "step": 17167
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.331763841758157e-05,
+      "loss": 0.0083,
+      "step": 17168
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.330181835462438e-05,
+      "loss": 0.0012,
+      "step": 17169
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.328599973969046e-05,
+      "loss": 0.0053,
+      "step": 17170
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.327018257304405e-05,
+      "loss": 0.0011,
+      "step": 17171
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.325436685494925e-05,
+      "loss": 0.0038,
+      "step": 17172
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.323855258567038e-05,
+      "loss": 0.0002,
+      "step": 17173
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.322273976547139e-05,
+      "loss": 0.0127,
+      "step": 17174
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.320692839461645e-05,
+      "loss": 0.002,
+      "step": 17175
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.319111847336965e-05,
+      "loss": 0.0006,
+      "step": 17176
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.31753100019951e-05,
+      "loss": 0.0017,
+      "step": 17177
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.315950298075662e-05,
+      "loss": 0.0027,
+      "step": 17178
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.314369740991843e-05,
+      "loss": 0.0075,
+      "step": 17179
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.312789328974446e-05,
+      "loss": 0.0007,
+      "step": 17180
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.311209062049866e-05,
+      "loss": 0.0016,
+      "step": 17181
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.309628940244498e-05,
+      "loss": 0.0086,
+      "step": 17182
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.308048963584725e-05,
+      "loss": 0.0005,
+      "step": 17183
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.306469132096947e-05,
+      "loss": 0.0086,
+      "step": 17184
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.304889445807537e-05,
+      "loss": 0.0012,
+      "step": 17185
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.303309904742898e-05,
+      "loss": 0.0107,
+      "step": 17186
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.301730508929393e-05,
+      "loss": 0.009,
+      "step": 17187
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.300151258393407e-05,
+      "loss": 0.0058,
+      "step": 17188
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.298572153161318e-05,
+      "loss": 0.0084,
+      "step": 17189
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.296993193259499e-05,
+      "loss": 0.0005,
+      "step": 17190
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.295414378714322e-05,
+      "loss": 0.0066,
+      "step": 17191
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.293835709552154e-05,
+      "loss": 0.0024,
+      "step": 17192
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.292257185799366e-05,
+      "loss": 0.0009,
+      "step": 17193
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.29067880748232e-05,
+      "loss": 0.0126,
+      "step": 17194
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.289100574627378e-05,
+      "loss": 0.0072,
+      "step": 17195
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.287522487260897e-05,
+      "loss": 0.0222,
+      "step": 17196
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.285944545409245e-05,
+      "loss": 0.0063,
+      "step": 17197
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.284366749098755e-05,
+      "loss": 0.0102,
+      "step": 17198
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.282789098355805e-05,
+      "loss": 0.0011,
+      "step": 17199
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.281211593206724e-05,
+      "loss": 0.0109,
+      "step": 17200
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.27963423367787e-05,
+      "loss": 0.0052,
+      "step": 17201
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.278057019795585e-05,
+      "loss": 0.0163,
+      "step": 17202
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.276479951586204e-05,
+      "loss": 0.0029,
+      "step": 17203
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.274903029076086e-05,
+      "loss": 0.0029,
+      "step": 17204
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.273326252291553e-05,
+      "loss": 0.0021,
+      "step": 17205
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.271749621258943e-05,
+      "loss": 0.0042,
+      "step": 17206
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.27017313600459e-05,
+      "loss": 0.0037,
+      "step": 17207
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.268596796554825e-05,
+      "loss": 0.0028,
+      "step": 17208
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.267020602935976e-05,
+      "loss": 0.0098,
+      "step": 17209
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.265444555174367e-05,
+      "loss": 0.0012,
+      "step": 17210
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.26386865329632e-05,
+      "loss": 0.0011,
+      "step": 17211
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.262292897328158e-05,
+      "loss": 0.0072,
+      "step": 17212
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.2607172872962e-05,
+      "loss": 0.0029,
+      "step": 17213
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.259141823226758e-05,
+      "loss": 0.0021,
+      "step": 17214
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.257566505146154e-05,
+      "loss": 0.0021,
+      "step": 17215
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.25599133308068e-05,
+      "loss": 0.0015,
+      "step": 17216
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.254416307056666e-05,
+      "loss": 0.0016,
+      "step": 17217
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.252841427100398e-05,
+      "loss": 0.0025,
+      "step": 17218
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.251266693238202e-05,
+      "loss": 0.0032,
+      "step": 17219
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.24969210549636e-05,
+      "loss": 0.0009,
+      "step": 17220
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.248117663901178e-05,
+      "loss": 0.0051,
+      "step": 17221
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.24654336847895e-05,
+      "loss": 0.0065,
+      "step": 17222
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.244969219255972e-05,
+      "loss": 0.0019,
+      "step": 17223
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.243395216258534e-05,
+      "loss": 0.0008,
+      "step": 17224
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.241821359512925e-05,
+      "loss": 0.0033,
+      "step": 17225
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.240247649045433e-05,
+      "loss": 0.0093,
+      "step": 17226
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.238674084882338e-05,
+      "loss": 0.0057,
+      "step": 17227
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.237100667049931e-05,
+      "loss": 0.0073,
+      "step": 17228
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.235527395574472e-05,
+      "loss": 0.0012,
+      "step": 17229
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.233954270482261e-05,
+      "loss": 0.007,
+      "step": 17230
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.232381291799551e-05,
+      "loss": 0.0032,
+      "step": 17231
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.230808459552633e-05,
+      "loss": 0.0015,
+      "step": 17232
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.229235773767762e-05,
+      "loss": 0.0029,
+      "step": 17233
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.227663234471203e-05,
+      "loss": 0.001,
+      "step": 17234
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.226090841689239e-05,
+      "loss": 0.0067,
+      "step": 17235
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.22451859544811e-05,
+      "loss": 0.0006,
+      "step": 17236
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.222946495774093e-05,
+      "loss": 0.0004,
+      "step": 17237
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.221374542693435e-05,
+      "loss": 0.0013,
+      "step": 17238
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.219802736232392e-05,
+      "loss": 0.0033,
+      "step": 17239
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.218231076417216e-05,
+      "loss": 0.001,
+      "step": 17240
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.216659563274159e-05,
+      "loss": 0.0023,
+      "step": 17241
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.215088196829467e-05,
+      "loss": 0.0059,
+      "step": 17242
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.213516977109384e-05,
+      "loss": 0.0059,
+      "step": 17243
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.211945904140155e-05,
+      "loss": 0.0093,
+      "step": 17244
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.210374977948016e-05,
+      "loss": 0.0005,
+      "step": 17245
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.208804198559208e-05,
+      "loss": 0.0038,
+      "step": 17246
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.207233565999963e-05,
+      "loss": 0.0006,
+      "step": 17247
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.20566308029652e-05,
+      "loss": 0.0008,
+      "step": 17248
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.204092741475095e-05,
+      "loss": 0.0064,
+      "step": 17249
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.202522549561935e-05,
+      "loss": 0.0096,
+      "step": 17250
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.20095250458325e-05,
+      "loss": 0.009,
+      "step": 17251
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.199382606565266e-05,
+      "loss": 0.0065,
+      "step": 17252
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.197812855534205e-05,
+      "loss": 0.0057,
+      "step": 17253
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.196243251516286e-05,
+      "loss": 0.0004,
+      "step": 17254
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.194673794537722e-05,
+      "loss": 0.0039,
+      "step": 17255
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.193104484624727e-05,
+      "loss": 0.0025,
+      "step": 17256
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.191535321803511e-05,
+      "loss": 0.0074,
+      "step": 17257
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.189966306100282e-05,
+      "loss": 0.0072,
+      "step": 17258
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.188397437541247e-05,
+      "loss": 0.0004,
+      "step": 17259
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.186828716152607e-05,
+      "loss": 0.0076,
+      "step": 17260
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.185260141960569e-05,
+      "loss": 0.0035,
+      "step": 17261
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.183691714991315e-05,
+      "loss": 0.0031,
+      "step": 17262
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.182123435271058e-05,
+      "loss": 0.0083,
+      "step": 17263
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.180555302825982e-05,
+      "loss": 0.0109,
+      "step": 17264
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.178987317682283e-05,
+      "loss": 0.0219,
+      "step": 17265
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.17741947986615e-05,
+      "loss": 0.013,
+      "step": 17266
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.175851789403755e-05,
+      "loss": 0.0101,
+      "step": 17267
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.174284246321302e-05,
+      "loss": 0.0014,
+      "step": 17268
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.172716850644953e-05,
+      "loss": 0.005,
+      "step": 17269
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.171149602400905e-05,
+      "loss": 0.0008,
+      "step": 17270
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.169582501615318e-05,
+      "loss": 0.0003,
+      "step": 17271
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.168015548314373e-05,
+      "loss": 0.0038,
+      "step": 17272
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.166448742524239e-05,
+      "loss": 0.0066,
+      "step": 17273
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.164882084271088e-05,
+      "loss": 0.0018,
+      "step": 17274
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.16331557358108e-05,
+      "loss": 0.0012,
+      "step": 17275
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.161749210480384e-05,
+      "loss": 0.0014,
+      "step": 17276
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.160182994995161e-05,
+      "loss": 0.0012,
+      "step": 17277
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.158616927151567e-05,
+      "loss": 0.0101,
+      "step": 17278
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.157051006975764e-05,
+      "loss": 0.0023,
+      "step": 17279
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.155485234493892e-05,
+      "loss": 0.0037,
+      "step": 17280
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.153919609732122e-05,
+      "loss": 0.0024,
+      "step": 17281
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.152354132716583e-05,
+      "loss": 0.0044,
+      "step": 17282
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.15078880347344e-05,
+      "loss": 0.0009,
+      "step": 17283
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.149223622028824e-05,
+      "loss": 0.0069,
+      "step": 17284
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.147658588408881e-05,
+      "loss": 0.0075,
+      "step": 17285
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.146093702639747e-05,
+      "loss": 0.009,
+      "step": 17286
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.144528964747559e-05,
+      "loss": 0.0223,
+      "step": 17287
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.142964374758463e-05,
+      "loss": 0.003,
+      "step": 17288
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.141399932698574e-05,
+      "loss": 0.0069,
+      "step": 17289
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.139835638594029e-05,
+      "loss": 0.0027,
+      "step": 17290
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.138271492470952e-05,
+      "loss": 0.003,
+      "step": 17291
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.136707494355471e-05,
+      "loss": 0.0078,
+      "step": 17292
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.135143644273704e-05,
+      "loss": 0.0084,
+      "step": 17293
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.133579942251773e-05,
+      "loss": 0.0019,
+      "step": 17294
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.132016388315793e-05,
+      "loss": 0.0025,
+      "step": 17295
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.130452982491878e-05,
+      "loss": 0.0042,
+      "step": 17296
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.128889724806148e-05,
+      "loss": 0.0017,
+      "step": 17297
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.127326615284692e-05,
+      "loss": 0.009,
+      "step": 17298
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.125763653953643e-05,
+      "loss": 0.0028,
+      "step": 17299
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.124200840839079e-05,
+      "loss": 0.0021,
+      "step": 17300
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.122638175967127e-05,
+      "loss": 0.0083,
+      "step": 17301
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.121075659363868e-05,
+      "loss": 0.0003,
+      "step": 17302
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.119513291055408e-05,
+      "loss": 0.0103,
+      "step": 17303
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.117951071067837e-05,
+      "loss": 0.0113,
+      "step": 17304
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.116388999427252e-05,
+      "loss": 0.0111,
+      "step": 17305
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.114827076159737e-05,
+      "loss": 0.0034,
+      "step": 17306
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.113265301291383e-05,
+      "loss": 0.0051,
+      "step": 17307
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.111703674848273e-05,
+      "loss": 0.0032,
+      "step": 17308
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.11014219685649e-05,
+      "loss": 0.0048,
+      "step": 17309
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.108580867342112e-05,
+      "loss": 0.0034,
+      "step": 17310
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.107019686331219e-05,
+      "loss": 0.0014,
+      "step": 17311
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.105458653849888e-05,
+      "loss": 0.0034,
+      "step": 17312
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.103897769924176e-05,
+      "loss": 0.0021,
+      "step": 17313
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.1023370345801764e-05,
+      "loss": 0.0052,
+      "step": 17314
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.100776447843932e-05,
+      "loss": 0.0009,
+      "step": 17315
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0992160097415246e-05,
+      "loss": 0.003,
+      "step": 17316
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.097655720299016e-05,
+      "loss": 0.0033,
+      "step": 17317
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0960955795424515e-05,
+      "loss": 0.0018,
+      "step": 17318
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0945355874979096e-05,
+      "loss": 0.0018,
+      "step": 17319
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0929757441914215e-05,
+      "loss": 0.0104,
+      "step": 17320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.091416049649062e-05,
+      "loss": 0.0016,
+      "step": 17321
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.089856503896866e-05,
+      "loss": 0.0078,
+      "step": 17322
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.088297106960887e-05,
+      "loss": 0.0048,
+      "step": 17323
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.086737858867166e-05,
+      "loss": 0.0115,
+      "step": 17324
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.08517875964175e-05,
+      "loss": 0.0059,
+      "step": 17325
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.083619809310675e-05,
+      "loss": 0.005,
+      "step": 17326
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0820610078999794e-05,
+      "loss": 0.004,
+      "step": 17327
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.080502355435701e-05,
+      "loss": 0.0104,
+      "step": 17328
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.078943851943868e-05,
+      "loss": 0.0045,
+      "step": 17329
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.077385497450518e-05,
+      "loss": 0.0002,
+      "step": 17330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.075827291981663e-05,
+      "loss": 0.006,
+      "step": 17331
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.074269235563347e-05,
+      "loss": 0.0051,
+      "step": 17332
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.072711328221574e-05,
+      "loss": 0.005,
+      "step": 17333
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0711535699823844e-05,
+      "loss": 0.0103,
+      "step": 17334
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.069595960871778e-05,
+      "loss": 0.0056,
+      "step": 17335
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.068038500915776e-05,
+      "loss": 0.0047,
+      "step": 17336
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0664811901403923e-05,
+      "loss": 0.0165,
+      "step": 17337
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.064924028571636e-05,
+      "loss": 0.0011,
+      "step": 17338
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.063367016235514e-05,
+      "loss": 0.0043,
+      "step": 17339
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0618101531580325e-05,
+      "loss": 0.0082,
+      "step": 17340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.060253439365194e-05,
+      "loss": 0.0109,
+      "step": 17341
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.058696874882996e-05,
+      "loss": 0.0021,
+      "step": 17342
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0571404597374396e-05,
+      "loss": 0.007,
+      "step": 17343
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.055584193954519e-05,
+      "loss": 0.0062,
+      "step": 17344
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0540280775602305e-05,
+      "loss": 0.0018,
+      "step": 17345
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0524721105805485e-05,
+      "loss": 0.0066,
+      "step": 17346
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.050916293041478e-05,
+      "loss": 0.0007,
+      "step": 17347
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.049360624969003e-05,
+      "loss": 0.0095,
+      "step": 17348
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0478051063890915e-05,
+      "loss": 0.005,
+      "step": 17349
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.046249737327742e-05,
+      "loss": 0.0114,
+      "step": 17350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.044694517810915e-05,
+      "loss": 0.0069,
+      "step": 17351
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.043139447864602e-05,
+      "loss": 0.0003,
+      "step": 17352
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.041584527514764e-05,
+      "loss": 0.0068,
+      "step": 17353
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.040029756787375e-05,
+      "loss": 0.0108,
+      "step": 17354
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0384751357084e-05,
+      "loss": 0.0065,
+      "step": 17355
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.036920664303808e-05,
+      "loss": 0.0059,
+      "step": 17356
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0353663425995587e-05,
+      "loss": 0.0105,
+      "step": 17357
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0338121706216154e-05,
+      "loss": 0.0319,
+      "step": 17358
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.032258148395931e-05,
+      "loss": 0.0016,
+      "step": 17359
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.030704275948465e-05,
+      "loss": 0.009,
+      "step": 17360
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0291505533051695e-05,
+      "loss": 0.0014,
+      "step": 17361
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.027596980491992e-05,
+      "loss": 0.0029,
+      "step": 17362
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0260435575348856e-05,
+      "loss": 0.015,
+      "step": 17363
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.024490284459783e-05,
+      "loss": 0.0053,
+      "step": 17364
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.022937161292645e-05,
+      "loss": 0.0022,
+      "step": 17365
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.02138418805939e-05,
+      "loss": 0.0014,
+      "step": 17366
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0198313647859795e-05,
+      "loss": 0.0035,
+      "step": 17367
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0182786914983297e-05,
+      "loss": 0.0028,
+      "step": 17368
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0167261682223764e-05,
+      "loss": 0.0007,
+      "step": 17369
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.015173794984064e-05,
+      "loss": 0.0054,
+      "step": 17370
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.013621571809298e-05,
+      "loss": 0.0099,
+      "step": 17371
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.012069498724024e-05,
+      "loss": 0.0017,
+      "step": 17372
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0105175757541504e-05,
+      "loss": 0.0151,
+      "step": 17373
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.008965802925603e-05,
+      "loss": 0.0068,
+      "step": 17374
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.007414180264299e-05,
+      "loss": 0.0049,
+      "step": 17375
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.005862707796152e-05,
+      "loss": 0.0063,
+      "step": 17376
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.004311385547075e-05,
+      "loss": 0.0032,
+      "step": 17377
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0027602135429794e-05,
+      "loss": 0.0038,
+      "step": 17378
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.001209191809771e-05,
+      "loss": 0.0006,
+      "step": 17379
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.999658320373356e-05,
+      "loss": 0.0072,
+      "step": 17380
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9981075992596426e-05,
+      "loss": 0.0004,
+      "step": 17381
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.996557028494512e-05,
+      "loss": 0.0059,
+      "step": 17382
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.995006608103886e-05,
+      "loss": 0.0041,
+      "step": 17383
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.993456338113637e-05,
+      "loss": 0.0031,
+      "step": 17384
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.991906218549678e-05,
+      "loss": 0.0014,
+      "step": 17385
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.990356249437885e-05,
+      "loss": 0.0084,
+      "step": 17386
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.988806430804149e-05,
+      "loss": 0.003,
+      "step": 17387
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.987256762674357e-05,
+      "loss": 0.006,
+      "step": 17388
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9857072450743884e-05,
+      "loss": 0.0093,
+      "step": 17389
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.984157878030124e-05,
+      "loss": 0.0074,
+      "step": 17390
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9826086615674424e-05,
+      "loss": 0.0016,
+      "step": 17391
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.981059595712217e-05,
+      "loss": 0.0032,
+      "step": 17392
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.979510680490321e-05,
+      "loss": 0.0056,
+      "step": 17393
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.977961915927623e-05,
+      "loss": 0.0051,
+      "step": 17394
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9764133020499935e-05,
+      "loss": 0.0026,
+      "step": 17395
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9748648388832986e-05,
+      "loss": 0.0039,
+      "step": 17396
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.973316526453389e-05,
+      "loss": 0.0032,
+      "step": 17397
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9717683647861396e-05,
+      "loss": 0.0051,
+      "step": 17398
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9702203539073974e-05,
+      "loss": 0.0033,
+      "step": 17399
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9686724938430145e-05,
+      "loss": 0.0034,
+      "step": 17400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.96712478461886e-05,
+      "loss": 0.022,
+      "step": 17401
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.965577226260761e-05,
+      "loss": 0.0047,
+      "step": 17402
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.964029818794586e-05,
+      "loss": 0.0006,
+      "step": 17403
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.962482562246165e-05,
+      "loss": 0.0032,
+      "step": 17404
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9609354566413446e-05,
+      "loss": 0.0081,
+      "step": 17405
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.959388502005965e-05,
+      "loss": 0.0013,
+      "step": 17406
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.957841698365863e-05,
+      "loss": 0.0063,
+      "step": 17407
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.956295045746872e-05,
+      "loss": 0.0021,
+      "step": 17408
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.954748544174826e-05,
+      "loss": 0.0112,
+      "step": 17409
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9532021936755525e-05,
+      "loss": 0.0136,
+      "step": 17410
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9516559942748795e-05,
+      "loss": 0.0074,
+      "step": 17411
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9501099459986306e-05,
+      "loss": 0.0023,
+      "step": 17412
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9485640488726296e-05,
+      "loss": 0.0006,
+      "step": 17413
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9470183029227e-05,
+      "loss": 0.0047,
+      "step": 17414
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.945472708174643e-05,
+      "loss": 0.0009,
+      "step": 17415
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.943927264654295e-05,
+      "loss": 0.0035,
+      "step": 17416
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.942381972387443e-05,
+      "loss": 0.0051,
+      "step": 17417
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.940836831399922e-05,
+      "loss": 0.0007,
+      "step": 17418
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9392918417175207e-05,
+      "loss": 0.0005,
+      "step": 17419
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9377470033660476e-05,
+      "loss": 0.0051,
+      "step": 17420
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.936202316371308e-05,
+      "loss": 0.0109,
+      "step": 17421
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.934657780759097e-05,
+      "loss": 0.0081,
+      "step": 17422
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9331133965552144e-05,
+      "loss": 0.0005,
+      "step": 17423
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.931569163785451e-05,
+      "loss": 0.003,
+      "step": 17424
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9300250824756023e-05,
+      "loss": 0.0036,
+      "step": 17425
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.928481152651456e-05,
+      "loss": 0.0107,
+      "step": 17426
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.926937374338803e-05,
+      "loss": 0.0008,
+      "step": 17427
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9253937475634106e-05,
+      "loss": 0.0047,
+      "step": 17428
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.923850272351079e-05,
+      "loss": 0.0071,
+      "step": 17429
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.922306948727579e-05,
+      "loss": 0.0193,
+      "step": 17430
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.920763776718689e-05,
+      "loss": 0.0044,
+      "step": 17431
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.919220756350187e-05,
+      "loss": 0.0017,
+      "step": 17432
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.917677887647829e-05,
+      "loss": 0.0054,
+      "step": 17433
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.916135170637406e-05,
+      "loss": 0.0071,
+      "step": 17434
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.914592605344662e-05,
+      "loss": 0.0026,
+      "step": 17435
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.913050191795381e-05,
+      "loss": 0.0068,
+      "step": 17436
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.911507930015309e-05,
+      "loss": 0.0009,
+      "step": 17437
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.909965820030212e-05,
+      "loss": 0.0039,
+      "step": 17438
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.908423861865843e-05,
+      "loss": 0.0124,
+      "step": 17439
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9068820555479586e-05,
+      "loss": 0.0037,
+      "step": 17440
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.905340401102306e-05,
+      "loss": 0.0014,
+      "step": 17441
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9037988985546375e-05,
+      "loss": 0.0066,
+      "step": 17442
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9022575479306974e-05,
+      "loss": 0.003,
+      "step": 17443
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.900716349256231e-05,
+      "loss": 0.0032,
+      "step": 17444
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.899175302556975e-05,
+      "loss": 0.0069,
+      "step": 17445
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.897634407858672e-05,
+      "loss": 0.0082,
+      "step": 17446
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8960936651870625e-05,
+      "loss": 0.0018,
+      "step": 17447
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.894553074567862e-05,
+      "loss": 0.0055,
+      "step": 17448
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.893012636026823e-05,
+      "loss": 0.0164,
+      "step": 17449
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.891472349589659e-05,
+      "loss": 0.0003,
+      "step": 17450
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8899322152821e-05,
+      "loss": 0.0043,
+      "step": 17451
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8883922331298696e-05,
+      "loss": 0.0031,
+      "step": 17452
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8868524031586834e-05,
+      "loss": 0.008,
+      "step": 17453
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.885312725394274e-05,
+      "loss": 0.0024,
+      "step": 17454
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8837731998623424e-05,
+      "loss": 0.0002,
+      "step": 17455
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.882233826588606e-05,
+      "loss": 0.0082,
+      "step": 17456
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.880694605598777e-05,
+      "loss": 0.0098,
+      "step": 17457
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.879155536918561e-05,
+      "loss": 0.0008,
+      "step": 17458
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.877616620573665e-05,
+      "loss": 0.0071,
+      "step": 17459
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.87607785658979e-05,
+      "loss": 0.0045,
+      "step": 17460
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.874539244992639e-05,
+      "loss": 0.0016,
+      "step": 17461
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.873000785807908e-05,
+      "loss": 0.0038,
+      "step": 17462
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.871462479061291e-05,
+      "loss": 0.007,
+      "step": 17463
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.869924324778483e-05,
+      "loss": 0.003,
+      "step": 17464
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8683863229851777e-05,
+      "loss": 0.0006,
+      "step": 17465
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8668484737070465e-05,
+      "loss": 0.0065,
+      "step": 17466
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.865310776969798e-05,
+      "loss": 0.0018,
+      "step": 17467
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8637732327990914e-05,
+      "loss": 0.0054,
+      "step": 17468
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8622358412206275e-05,
+      "loss": 0.0028,
+      "step": 17469
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8606986022600694e-05,
+      "loss": 0.0031,
+      "step": 17470
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.859161515943097e-05,
+      "loss": 0.0047,
+      "step": 17471
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8576245822953806e-05,
+      "loss": 0.0017,
+      "step": 17472
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.856087801342592e-05,
+      "loss": 0.0069,
+      "step": 17473
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8545511731103966e-05,
+      "loss": 0.0099,
+      "step": 17474
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.85301469762446e-05,
+      "loss": 0.0022,
+      "step": 17475
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.851478374910446e-05,
+      "loss": 0.0053,
+      "step": 17476
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8499422049940085e-05,
+      "loss": 0.0014,
+      "step": 17477
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.848406187900816e-05,
+      "loss": 0.0033,
+      "step": 17478
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.846870323656503e-05,
+      "loss": 0.0071,
+      "step": 17479
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.845334612286746e-05,
+      "loss": 0.0067,
+      "step": 17480
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8437990538171705e-05,
+      "loss": 0.0115,
+      "step": 17481
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.842263648273439e-05,
+      "loss": 0.0024,
+      "step": 17482
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8407283956811954e-05,
+      "loss": 0.0029,
+      "step": 17483
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8391932960660676e-05,
+      "loss": 0.002,
+      "step": 17484
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8376583494537137e-05,
+      "loss": 0.0082,
+      "step": 17485
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.836123555869752e-05,
+      "loss": 0.0009,
+      "step": 17486
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.834588915339832e-05,
+      "loss": 0.0009,
+      "step": 17487
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.833054427889575e-05,
+      "loss": 0.0031,
+      "step": 17488
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.831520093544613e-05,
+      "loss": 0.0024,
+      "step": 17489
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8299859123305725e-05,
+      "loss": 0.0065,
+      "step": 17490
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.828451884273076e-05,
+      "loss": 0.0097,
+      "step": 17491
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8269180093977465e-05,
+      "loss": 0.0008,
+      "step": 17492
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8253842877302e-05,
+      "loss": 0.0016,
+      "step": 17493
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.823850719296056e-05,
+      "loss": 0.0017,
+      "step": 17494
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.822317304120926e-05,
+      "loss": 0.003,
+      "step": 17495
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.820784042230424e-05,
+      "loss": 0.0076,
+      "step": 17496
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.819250933650147e-05,
+      "loss": 0.0004,
+      "step": 17497
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8177179784057205e-05,
+      "loss": 0.0027,
+      "step": 17498
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.816185176522724e-05,
+      "loss": 0.0011,
+      "step": 17499
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8146525280267825e-05,
+      "loss": 0.0012,
+      "step": 17500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8131200329434786e-05,
+      "loss": 0.0035,
+      "step": 17501
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.811587691298408e-05,
+      "loss": 0.0003,
+      "step": 17502
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8100555031171694e-05,
+      "loss": 0.0009,
+      "step": 17503
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.808523468425352e-05,
+      "loss": 0.0074,
+      "step": 17504
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8069915872485426e-05,
+      "loss": 0.0022,
+      "step": 17505
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.805459859612326e-05,
+      "loss": 0.0039,
+      "step": 17506
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8039282855422845e-05,
+      "loss": 0.0061,
+      "step": 17507
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.802396865064e-05,
+      "loss": 0.0037,
+      "step": 17508
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.800865598203051e-05,
+      "loss": 0.0135,
+      "step": 17509
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.79933448498501e-05,
+      "loss": 0.0082,
+      "step": 17510
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7978035254354563e-05,
+      "loss": 0.0081,
+      "step": 17511
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.796272719579943e-05,
+      "loss": 0.002,
+      "step": 17512
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.794742067444055e-05,
+      "loss": 0.0023,
+      "step": 17513
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7932115690533495e-05,
+      "loss": 0.0063,
+      "step": 17514
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.791681224433389e-05,
+      "loss": 0.0016,
+      "step": 17515
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.790151033609742e-05,
+      "loss": 0.0117,
+      "step": 17516
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.788620996607944e-05,
+      "loss": 0.0101,
+      "step": 17517
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.787091113453575e-05,
+      "loss": 0.0064,
+      "step": 17518
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7855613841721655e-05,
+      "loss": 0.0051,
+      "step": 17519
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7840318087892846e-05,
+      "loss": 0.0008,
+      "step": 17520
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.782502387330462e-05,
+      "loss": 0.0066,
+      "step": 17521
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7809731198212494e-05,
+      "loss": 0.0034,
+      "step": 17522
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7794440062871894e-05,
+      "loss": 0.0021,
+      "step": 17523
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.777915046753818e-05,
+      "loss": 0.0057,
+      "step": 17524
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.776386241246672e-05,
+      "loss": 0.0044,
+      "step": 17525
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.774857589791289e-05,
+      "loss": 0.0133,
+      "step": 17526
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.773329092413197e-05,
+      "loss": 0.0047,
+      "step": 17527
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7718007491379254e-05,
+      "loss": 0.0007,
+      "step": 17528
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.770272559991007e-05,
+      "loss": 0.004,
+      "step": 17529
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.768744524997948e-05,
+      "loss": 0.0053,
+      "step": 17530
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.767216644184291e-05,
+      "loss": 0.0032,
+      "step": 17531
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7656889175755335e-05,
+      "loss": 0.0071,
+      "step": 17532
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7641613451972135e-05,
+      "loss": 0.0035,
+      "step": 17533
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.762633927074826e-05,
+      "loss": 0.0018,
+      "step": 17534
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.761106663233885e-05,
+      "loss": 0.009,
+      "step": 17535
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7595795536999124e-05,
+      "loss": 0.0028,
+      "step": 17536
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7580525984983935e-05,
+      "loss": 0.0014,
+      "step": 17537
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7565257976548525e-05,
+      "loss": 0.0008,
+      "step": 17538
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.754999151194772e-05,
+      "loss": 0.0147,
+      "step": 17539
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.753472659143658e-05,
+      "loss": 0.0086,
+      "step": 17540
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.751946321527006e-05,
+      "loss": 0.0071,
+      "step": 17541
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7504201383703045e-05,
+      "loss": 0.0118,
+      "step": 17542
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.748894109699048e-05,
+      "loss": 0.0085,
+      "step": 17543
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.747368235538724e-05,
+      "loss": 0.0009,
+      "step": 17544
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.745842515914813e-05,
+      "loss": 0.0064,
+      "step": 17545
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.744316950852803e-05,
+      "loss": 0.0194,
+      "step": 17546
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.742791540378175e-05,
+      "loss": 0.0002,
+      "step": 17547
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7412662845163935e-05,
+      "loss": 0.002,
+      "step": 17548
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.739741183292951e-05,
+      "loss": 0.0005,
+      "step": 17549
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.738216236733301e-05,
+      "loss": 0.0037,
+      "step": 17550
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.736691444862933e-05,
+      "loss": 0.0023,
+      "step": 17551
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.735166807707299e-05,
+      "loss": 0.006,
+      "step": 17552
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7336423252918676e-05,
+      "loss": 0.0105,
+      "step": 17553
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7321179976421e-05,
+      "loss": 0.0004,
+      "step": 17554
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.730593824783457e-05,
+      "loss": 0.0067,
+      "step": 17555
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7290698067413954e-05,
+      "loss": 0.003,
+      "step": 17556
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.727545943541366e-05,
+      "loss": 0.0074,
+      "step": 17557
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.726022235208824e-05,
+      "loss": 0.0014,
+      "step": 17558
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.724498681769216e-05,
+      "loss": 0.0034,
+      "step": 17559
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7229752832479885e-05,
+      "loss": 0.0044,
+      "step": 17560
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.721452039670586e-05,
+      "loss": 0.0057,
+      "step": 17561
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7199289510624525e-05,
+      "loss": 0.006,
+      "step": 17562
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.718406017449014e-05,
+      "loss": 0.0009,
+      "step": 17563
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7168832388557264e-05,
+      "loss": 0.0034,
+      "step": 17564
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.715360615307999e-05,
+      "loss": 0.0065,
+      "step": 17565
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7138381468312824e-05,
+      "loss": 0.0031,
+      "step": 17566
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.712315833451002e-05,
+      "loss": 0.0026,
+      "step": 17567
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.710793675192569e-05,
+      "loss": 0.0005,
+      "step": 17568
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7092716720814264e-05,
+      "loss": 0.0057,
+      "step": 17569
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.70774982414298e-05,
+      "loss": 0.0019,
+      "step": 17570
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.706228131402651e-05,
+      "loss": 0.0021,
+      "step": 17571
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7047065938858545e-05,
+      "loss": 0.0013,
+      "step": 17572
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.703185211618005e-05,
+      "loss": 0.0037,
+      "step": 17573
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7016639846245124e-05,
+      "loss": 0.0005,
+      "step": 17574
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7001429129307825e-05,
+      "loss": 0.0066,
+      "step": 17575
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.698621996562221e-05,
+      "loss": 0.0046,
+      "step": 17576
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.697101235544227e-05,
+      "loss": 0.0041,
+      "step": 17577
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.695580629902207e-05,
+      "loss": 0.0019,
+      "step": 17578
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.694060179661551e-05,
+      "loss": 0.0083,
+      "step": 17579
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.692539884847661e-05,
+      "loss": 0.0097,
+      "step": 17580
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6910197454859154e-05,
+      "loss": 0.0028,
+      "step": 17581
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.689499761601723e-05,
+      "loss": 0.0035,
+      "step": 17582
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.687979933220448e-05,
+      "loss": 0.0006,
+      "step": 17583
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.686460260367495e-05,
+      "loss": 0.0043,
+      "step": 17584
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6849407430682315e-05,
+      "loss": 0.0068,
+      "step": 17585
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.683421381348043e-05,
+      "loss": 0.0076,
+      "step": 17586
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.681902175232302e-05,
+      "loss": 0.0033,
+      "step": 17587
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6803831247463784e-05,
+      "loss": 0.0006,
+      "step": 17588
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.67886422991566e-05,
+      "loss": 0.0114,
+      "step": 17589
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.677345490765497e-05,
+      "loss": 0.0097,
+      "step": 17590
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.675826907321264e-05,
+      "loss": 0.0025,
+      "step": 17591
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6743084796083216e-05,
+      "loss": 0.0114,
+      "step": 17592
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.67279020765203e-05,
+      "loss": 0.0214,
+      "step": 17593
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.67127209147775e-05,
+      "loss": 0.0036,
+      "step": 17594
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.669754131110835e-05,
+      "loss": 0.0002,
+      "step": 17595
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6682363265766376e-05,
+      "loss": 0.0112,
+      "step": 17596
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6667186779005075e-05,
+      "loss": 0.0019,
+      "step": 17597
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6652011851078e-05,
+      "loss": 0.0112,
+      "step": 17598
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.663683848223843e-05,
+      "loss": 0.009,
+      "step": 17599
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.662166667273999e-05,
+      "loss": 0.0052,
+      "step": 17600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.660649642283588e-05,
+      "loss": 0.0039,
+      "step": 17601
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.659132773277968e-05,
+      "loss": 0.0089,
+      "step": 17602
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.657616060282457e-05,
+      "loss": 0.0032,
+      "step": 17603
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.656099503322393e-05,
+      "loss": 0.015,
+      "step": 17604
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.654583102423104e-05,
+      "loss": 0.0073,
+      "step": 17605
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6530668576099196e-05,
+      "loss": 0.0038,
+      "step": 17606
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.651550768908162e-05,
+      "loss": 0.0038,
+      "step": 17607
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.650034836343152e-05,
+      "loss": 0.0095,
+      "step": 17608
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.648519059940211e-05,
+      "loss": 0.0018,
+      "step": 17609
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.647003439724652e-05,
+      "loss": 0.0067,
+      "step": 17610
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.645487975721793e-05,
+      "loss": 0.0124,
+      "step": 17611
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.64397266795694e-05,
+      "loss": 0.0184,
+      "step": 17612
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.642457516455412e-05,
+      "loss": 0.0043,
+      "step": 17613
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.640942521242496e-05,
+      "loss": 0.0128,
+      "step": 17614
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.639427682343515e-05,
+      "loss": 0.0251,
+      "step": 17615
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.637912999783752e-05,
+      "loss": 0.0005,
+      "step": 17616
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.636398473588525e-05,
+      "loss": 0.0026,
+      "step": 17617
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.634884103783113e-05,
+      "loss": 0.0014,
+      "step": 17618
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.633369890392809e-05,
+      "loss": 0.0059,
+      "step": 17619
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.631855833442918e-05,
+      "loss": 0.003,
+      "step": 17620
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6303419329587146e-05,
+      "loss": 0.0019,
+      "step": 17621
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.628828188965486e-05,
+      "loss": 0.0016,
+      "step": 17622
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.627314601488516e-05,
+      "loss": 0.0007,
+      "step": 17623
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.625801170553086e-05,
+      "loss": 0.005,
+      "step": 17624
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.62428789618447e-05,
+      "loss": 0.0044,
+      "step": 17625
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.622774778407945e-05,
+      "loss": 0.0017,
+      "step": 17626
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.621261817248781e-05,
+      "loss": 0.0121,
+      "step": 17627
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6197490127322494e-05,
+      "loss": 0.0067,
+      "step": 17628
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.618236364883616e-05,
+      "loss": 0.001,
+      "step": 17629
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.616723873728143e-05,
+      "loss": 0.0026,
+      "step": 17630
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.615211539291099e-05,
+      "loss": 0.0016,
+      "step": 17631
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6136993615977276e-05,
+      "loss": 0.0015,
+      "step": 17632
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.612187340673305e-05,
+      "loss": 0.0066,
+      "step": 17633
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.610675476543066e-05,
+      "loss": 0.0139,
+      "step": 17634
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6091637692322784e-05,
+      "loss": 0.0023,
+      "step": 17635
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.607652218766177e-05,
+      "loss": 0.0034,
+      "step": 17636
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6061408251700137e-05,
+      "loss": 0.0087,
+      "step": 17637
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.604629588469029e-05,
+      "loss": 0.0003,
+      "step": 17638
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.603118508688468e-05,
+      "loss": 0.0015,
+      "step": 17639
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.601607585853563e-05,
+      "loss": 0.0016,
+      "step": 17640
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.600096819989555e-05,
+      "loss": 0.0115,
+      "step": 17641
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5985862111216705e-05,
+      "loss": 0.0036,
+      "step": 17642
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.597075759275144e-05,
+      "loss": 0.0084,
+      "step": 17643
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5955654644752085e-05,
+      "loss": 0.0033,
+      "step": 17644
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.59405532674707e-05,
+      "loss": 0.0055,
+      "step": 17645
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.592545346115972e-05,
+      "loss": 0.0003,
+      "step": 17646
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.591035522607116e-05,
+      "loss": 0.0041,
+      "step": 17647
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.589525856245731e-05,
+      "loss": 0.0018,
+      "step": 17648
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.588016347057033e-05,
+      "loss": 0.0007,
+      "step": 17649
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.58650699506622e-05,
+      "loss": 0.0081,
+      "step": 17650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5849978002985177e-05,
+      "loss": 0.0018,
+      "step": 17651
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5834887627791155e-05,
+      "loss": 0.0112,
+      "step": 17652
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5819798825332354e-05,
+      "loss": 0.0079,
+      "step": 17653
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.580471159586064e-05,
+      "loss": 0.0016,
+      "step": 17654
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5789625939628035e-05,
+      "loss": 0.0022,
+      "step": 17655
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5774541856886534e-05,
+      "loss": 0.0106,
+      "step": 17656
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5759459347888034e-05,
+      "loss": 0.0048,
+      "step": 17657
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5744378412884456e-05,
+      "loss": 0.0079,
+      "step": 17658
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5729299052127674e-05,
+      "loss": 0.001,
+      "step": 17659
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5714221265869554e-05,
+      "loss": 0.0059,
+      "step": 17660
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5699145054361924e-05,
+      "loss": 0.002,
+      "step": 17661
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.568407041785656e-05,
+      "loss": 0.0011,
+      "step": 17662
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5668997356605265e-05,
+      "loss": 0.0054,
+      "step": 17663
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.565392587085983e-05,
+      "loss": 0.0051,
+      "step": 17664
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.56388559608718e-05,
+      "loss": 0.007,
+      "step": 17665
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.562378762689312e-05,
+      "loss": 0.0111,
+      "step": 17666
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5608720869175225e-05,
+      "loss": 0.0007,
+      "step": 17667
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.559365568796997e-05,
+      "loss": 0.0083,
+      "step": 17668
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5578592083528814e-05,
+      "loss": 0.0199,
+      "step": 17669
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.55635300561034e-05,
+      "loss": 0.0029,
+      "step": 17670
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.554846960594531e-05,
+      "loss": 0.0016,
+      "step": 17671
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.553341073330606e-05,
+      "loss": 0.0097,
+      "step": 17672
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.551835343843716e-05,
+      "loss": 0.0094,
+      "step": 17673
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.550329772159011e-05,
+      "loss": 0.0024,
+      "step": 17674
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5488243583016375e-05,
+      "loss": 0.0007,
+      "step": 17675
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5473191022967364e-05,
+      "loss": 0.0007,
+      "step": 17676
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.545814004169456e-05,
+      "loss": 0.0043,
+      "step": 17677
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.544309063944915e-05,
+      "loss": 0.0066,
+      "step": 17678
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5428042816482685e-05,
+      "loss": 0.0004,
+      "step": 17679
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.541299657304642e-05,
+      "loss": 0.0008,
+      "step": 17680
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.539795190939166e-05,
+      "loss": 0.0005,
+      "step": 17681
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5382908825769727e-05,
+      "loss": 0.0044,
+      "step": 17682
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5367867322431716e-05,
+      "loss": 0.0118,
+      "step": 17683
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.535282739962905e-05,
+      "loss": 0.0013,
+      "step": 17684
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.533778905761272e-05,
+      "loss": 0.0065,
+      "step": 17685
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5322752296634117e-05,
+      "loss": 0.0003,
+      "step": 17686
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.530771711694421e-05,
+      "loss": 0.0016,
+      "step": 17687
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.529268351879416e-05,
+      "loss": 0.0095,
+      "step": 17688
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.527765150243507e-05,
+      "loss": 0.0092,
+      "step": 17689
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5262621068118e-05,
+      "loss": 0.0003,
+      "step": 17690
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5247592216094e-05,
+      "loss": 0.0012,
+      "step": 17691
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.523256494661406e-05,
+      "loss": 0.0015,
+      "step": 17692
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.521753925992918e-05,
+      "loss": 0.0104,
+      "step": 17693
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.52025151562903e-05,
+      "loss": 0.0011,
+      "step": 17694
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.518749263594841e-05,
+      "loss": 0.0018,
+      "step": 17695
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5172471699154264e-05,
+      "loss": 0.0049,
+      "step": 17696
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5157452346158944e-05,
+      "loss": 0.0092,
+      "step": 17697
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.51424345772131e-05,
+      "loss": 0.0056,
+      "step": 17698
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.512741839256776e-05,
+      "loss": 0.0207,
+      "step": 17699
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5112403792473555e-05,
+      "loss": 0.0123,
+      "step": 17700
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.509739077718127e-05,
+      "loss": 0.0006,
+      "step": 17701
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.508237934694181e-05,
+      "loss": 0.0024,
+      "step": 17702
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.506736950200568e-05,
+      "loss": 0.001,
+      "step": 17703
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5052361242623773e-05,
+      "loss": 0.004,
+      "step": 17704
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.503735456904662e-05,
+      "loss": 0.0023,
+      "step": 17705
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.502234948152488e-05,
+      "loss": 0.0101,
+      "step": 17706
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.50073459803092e-05,
+      "loss": 0.0019,
+      "step": 17707
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.499234406565015e-05,
+      "loss": 0.0034,
+      "step": 17708
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.497734373779831e-05,
+      "loss": 0.0047,
+      "step": 17709
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4962344997004184e-05,
+      "loss": 0.0004,
+      "step": 17710
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.494734784351829e-05,
+      "loss": 0.0013,
+      "step": 17711
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.493235227759112e-05,
+      "loss": 0.0069,
+      "step": 17712
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4917358299473115e-05,
+      "loss": 0.0021,
+      "step": 17713
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.490236590941471e-05,
+      "loss": 0.0031,
+      "step": 17714
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4887375107666355e-05,
+      "loss": 0.0005,
+      "step": 17715
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4872385894478263e-05,
+      "loss": 0.0037,
+      "step": 17716
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4857398270100994e-05,
+      "loss": 0.006,
+      "step": 17717
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.484241223478472e-05,
+      "loss": 0.0013,
+      "step": 17718
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.48274277887798e-05,
+      "loss": 0.0107,
+      "step": 17719
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4812444932336474e-05,
+      "loss": 0.0141,
+      "step": 17720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4797463665704996e-05,
+      "loss": 0.0003,
+      "step": 17721
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.478248398913559e-05,
+      "loss": 0.0089,
+      "step": 17722
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4767505902878427e-05,
+      "loss": 0.0056,
+      "step": 17723
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.475252940718368e-05,
+      "loss": 0.0155,
+      "step": 17724
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4737554502301485e-05,
+      "loss": 0.0045,
+      "step": 17725
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.472258118848195e-05,
+      "loss": 0.0033,
+      "step": 17726
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.470760946597517e-05,
+      "loss": 0.0076,
+      "step": 17727
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.469263933503123e-05,
+      "loss": 0.0031,
+      "step": 17728
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4677670795900026e-05,
+      "loss": 0.0105,
+      "step": 17729
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4662703848831736e-05,
+      "loss": 0.0029,
+      "step": 17730
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4647738494076174e-05,
+      "loss": 0.0015,
+      "step": 17731
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.46327747318834e-05,
+      "loss": 0.0064,
+      "step": 17732
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.461781256250337e-05,
+      "loss": 0.0085,
+      "step": 17733
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.460285198618581e-05,
+      "loss": 0.0056,
+      "step": 17734
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.458789300318081e-05,
+      "loss": 0.0118,
+      "step": 17735
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.457293561373798e-05,
+      "loss": 0.0008,
+      "step": 17736
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.455797981810736e-05,
+      "loss": 0.0041,
+      "step": 17737
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.45430256165386e-05,
+      "loss": 0.0074,
+      "step": 17738
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.452807300928149e-05,
+      "loss": 0.0059,
+      "step": 17739
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.451312199658579e-05,
+      "loss": 0.0051,
+      "step": 17740
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4498172578701186e-05,
+      "loss": 0.0051,
+      "step": 17741
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.448322475587737e-05,
+      "loss": 0.0012,
+      "step": 17742
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4468278528364e-05,
+      "loss": 0.0084,
+      "step": 17743
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.445333389641072e-05,
+      "loss": 0.0046,
+      "step": 17744
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.44383908602671e-05,
+      "loss": 0.0058,
+      "step": 17745
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.44234494201828e-05,
+      "loss": 0.0009,
+      "step": 17746
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4408509576407205e-05,
+      "loss": 0.0025,
+      "step": 17747
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.439357132919006e-05,
+      "loss": 0.0032,
+      "step": 17748
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.437863467878062e-05,
+      "loss": 0.0025,
+      "step": 17749
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4363699625428596e-05,
+      "loss": 0.0121,
+      "step": 17750
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.434876616938324e-05,
+      "loss": 0.0042,
+      "step": 17751
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.433383431089408e-05,
+      "loss": 0.0043,
+      "step": 17752
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4318904050210436e-05,
+      "loss": 0.0083,
+      "step": 17753
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.430397538758168e-05,
+      "loss": 0.0047,
+      "step": 17754
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.428904832325727e-05,
+      "loss": 0.001,
+      "step": 17755
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.427412285748636e-05,
+      "loss": 0.0006,
+      "step": 17756
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.425919899051829e-05,
+      "loss": 0.0071,
+      "step": 17757
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.424427672260232e-05,
+      "loss": 0.003,
+      "step": 17758
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.422935605398769e-05,
+      "loss": 0.0066,
+      "step": 17759
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4214436984923595e-05,
+      "loss": 0.0052,
+      "step": 17760
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.419951951565926e-05,
+      "loss": 0.0021,
+      "step": 17761
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.418460364644368e-05,
+      "loss": 0.0016,
+      "step": 17762
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.416968937752613e-05,
+      "loss": 0.0193,
+      "step": 17763
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.415477670915568e-05,
+      "loss": 0.0142,
+      "step": 17764
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4139865641581356e-05,
+      "loss": 0.0022,
+      "step": 17765
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4124956175052295e-05,
+      "loss": 0.0011,
+      "step": 17766
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.411004830981733e-05,
+      "loss": 0.0095,
+      "step": 17767
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.409514204612566e-05,
+      "loss": 0.0074,
+      "step": 17768
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.408023738422611e-05,
+      "loss": 0.0095,
+      "step": 17769
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.406533432436764e-05,
+      "loss": 0.0023,
+      "step": 17770
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4050432866799194e-05,
+      "loss": 0.0046,
+      "step": 17771
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.403553301176964e-05,
+      "loss": 0.0026,
+      "step": 17772
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.402063475952781e-05,
+      "loss": 0.0065,
+      "step": 17773
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.400573811032257e-05,
+      "loss": 0.0041,
+      "step": 17774
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3990843064402716e-05,
+      "loss": 0.005,
+      "step": 17775
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3975949622017e-05,
+      "loss": 0.0022,
+      "step": 17776
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3961057783414176e-05,
+      "loss": 0.0002,
+      "step": 17777
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.394616754884297e-05,
+      "loss": 0.0117,
+      "step": 17778
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.393127891855214e-05,
+      "loss": 0.0009,
+      "step": 17779
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.391639189279021e-05,
+      "loss": 0.0027,
+      "step": 17780
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3901506471805987e-05,
+      "loss": 0.0031,
+      "step": 17781
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.388662265584791e-05,
+      "loss": 0.0014,
+      "step": 17782
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.387174044516476e-05,
+      "loss": 0.0014,
+      "step": 17783
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.385685984000494e-05,
+      "loss": 0.0019,
+      "step": 17784
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.384198084061701e-05,
+      "loss": 0.0004,
+      "step": 17785
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.382710344724961e-05,
+      "loss": 0.0034,
+      "step": 17786
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.381222766015101e-05,
+      "loss": 0.0034,
+      "step": 17787
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3797353479569874e-05,
+      "loss": 0.0016,
+      "step": 17788
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3782480905754496e-05,
+      "loss": 0.0007,
+      "step": 17789
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3767609938953296e-05,
+      "loss": 0.0033,
+      "step": 17790
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.375274057941468e-05,
+      "loss": 0.0035,
+      "step": 17791
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.373787282738695e-05,
+      "loss": 0.0009,
+      "step": 17792
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.372300668311847e-05,
+      "loss": 0.005,
+      "step": 17793
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.370814214685751e-05,
+      "loss": 0.0013,
+      "step": 17794
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.369327921885233e-05,
+      "loss": 0.0043,
+      "step": 17795
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.367841789935119e-05,
+      "loss": 0.0038,
+      "step": 17796
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.366355818860234e-05,
+      "loss": 0.0018,
+      "step": 17797
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3648700086853823e-05,
+      "loss": 0.0125,
+      "step": 17798
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.363384359435398e-05,
+      "loss": 0.006,
+      "step": 17799
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.361898871135075e-05,
+      "loss": 0.0123,
+      "step": 17800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3604135438092453e-05,
+      "loss": 0.0053,
+      "step": 17801
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3589283774827e-05,
+      "loss": 0.0046,
+      "step": 17802
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.357443372180249e-05,
+      "loss": 0.0078,
+      "step": 17803
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.355958527926695e-05,
+      "loss": 0.0058,
+      "step": 17804
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.35447384474684e-05,
+      "loss": 0.0057,
+      "step": 17805
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3529893226654766e-05,
+      "loss": 0.005,
+      "step": 17806
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.351504961707403e-05,
+      "loss": 0.0031,
+      "step": 17807
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3500207618974075e-05,
+      "loss": 0.0069,
+      "step": 17808
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.348536723260283e-05,
+      "loss": 0.0019,
+      "step": 17809
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.347052845820813e-05,
+      "loss": 0.0002,
+      "step": 17810
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.345569129603782e-05,
+      "loss": 0.0019,
+      "step": 17811
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.344085574633976e-05,
+      "loss": 0.0133,
+      "step": 17812
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.342602180936156e-05,
+      "loss": 0.0098,
+      "step": 17813
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3411189485351205e-05,
+      "loss": 0.0047,
+      "step": 17814
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3396358774556215e-05,
+      "loss": 0.0139,
+      "step": 17815
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.338152967722441e-05,
+      "loss": 0.0014,
+      "step": 17816
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.336670219360352e-05,
+      "loss": 0.0045,
+      "step": 17817
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3351876323941e-05,
+      "loss": 0.0006,
+      "step": 17818
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.333705206848467e-05,
+      "loss": 0.0011,
+      "step": 17819
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.332222942748199e-05,
+      "loss": 0.0064,
+      "step": 17820
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.330740840118059e-05,
+      "loss": 0.0046,
+      "step": 17821
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3292588989827976e-05,
+      "loss": 0.0015,
+      "step": 17822
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.327777119367168e-05,
+      "loss": 0.0021,
+      "step": 17823
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.326295501295917e-05,
+      "loss": 0.0062,
+      "step": 17824
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.324814044793792e-05,
+      "loss": 0.0014,
+      "step": 17825
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.323332749885535e-05,
+      "loss": 0.0163,
+      "step": 17826
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.321851616595889e-05,
+      "loss": 0.0002,
+      "step": 17827
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3203706449495885e-05,
+      "loss": 0.0041,
+      "step": 17828
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.318889834971369e-05,
+      "loss": 0.0003,
+      "step": 17829
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3174091866859697e-05,
+      "loss": 0.0066,
+      "step": 17830
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.315928700118105e-05,
+      "loss": 0.0132,
+      "step": 17831
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.31444837529252e-05,
+      "loss": 0.0135,
+      "step": 17832
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3129682122339196e-05,
+      "loss": 0.0035,
+      "step": 17833
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.311488210967046e-05,
+      "loss": 0.0047,
+      "step": 17834
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.310008371516603e-05,
+      "loss": 0.0011,
+      "step": 17835
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.30852869390731e-05,
+      "loss": 0.0022,
+      "step": 17836
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.307049178163883e-05,
+      "loss": 0.0097,
+      "step": 17837
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.305569824311026e-05,
+      "loss": 0.0034,
+      "step": 17838
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.304090632373463e-05,
+      "loss": 0.0014,
+      "step": 17839
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.302611602375883e-05,
+      "loss": 0.0011,
+      "step": 17840
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3011327343429955e-05,
+      "loss": 0.0041,
+      "step": 17841
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.299654028299497e-05,
+      "loss": 0.0037,
+      "step": 17842
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.298175484270094e-05,
+      "loss": 0.0033,
+      "step": 17843
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2966971022794636e-05,
+      "loss": 0.0021,
+      "step": 17844
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2952188823523126e-05,
+      "loss": 0.0062,
+      "step": 17845
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2937408245133254e-05,
+      "loss": 0.0053,
+      "step": 17846
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2922629287871885e-05,
+      "loss": 0.0081,
+      "step": 17847
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.29078519519859e-05,
+      "loss": 0.0017,
+      "step": 17848
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2893076237721956e-05,
+      "loss": 0.0052,
+      "step": 17849
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2878302145327046e-05,
+      "loss": 0.0067,
+      "step": 17850
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.286352967504773e-05,
+      "loss": 0.0053,
+      "step": 17851
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.284875882713092e-05,
+      "loss": 0.0053,
+      "step": 17852
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.283398960182316e-05,
+      "loss": 0.0032,
+      "step": 17853
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.28192219993712e-05,
+      "loss": 0.0029,
+      "step": 17854
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.280445602002168e-05,
+      "loss": 0.0026,
+      "step": 17855
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.278969166402121e-05,
+      "loss": 0.0013,
+      "step": 17856
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.277492893161638e-05,
+      "loss": 0.0013,
+      "step": 17857
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.276016782305377e-05,
+      "loss": 0.0013,
+      "step": 17858
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.274540833857991e-05,
+      "loss": 0.0085,
+      "step": 17859
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.273065047844131e-05,
+      "loss": 0.005,
+      "step": 17860
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.271589424288446e-05,
+      "loss": 0.004,
+      "step": 17861
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2701139632155796e-05,
+      "loss": 0.0036,
+      "step": 17862
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2686386646501827e-05,
+      "loss": 0.0074,
+      "step": 17863
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.26716352861688e-05,
+      "loss": 0.0005,
+      "step": 17864
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.265688555140327e-05,
+      "loss": 0.0072,
+      "step": 17865
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.264213744245146e-05,
+      "loss": 0.0026,
+      "step": 17866
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.262739095955973e-05,
+      "loss": 0.0011,
+      "step": 17867
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.261264610297438e-05,
+      "loss": 0.0094,
+      "step": 17868
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2597902872941604e-05,
+      "loss": 0.003,
+      "step": 17869
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.258316126970782e-05,
+      "loss": 0.0007,
+      "step": 17870
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.256842129351909e-05,
+      "loss": 0.0026,
+      "step": 17871
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.255368294462163e-05,
+      "loss": 0.0011,
+      "step": 17872
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.25389462232616e-05,
+      "loss": 0.0048,
+      "step": 17873
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.252421112968515e-05,
+      "loss": 0.0057,
+      "step": 17874
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.250947766413836e-05,
+      "loss": 0.0069,
+      "step": 17875
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2494745826867336e-05,
+      "loss": 0.01,
+      "step": 17876
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.248001561811809e-05,
+      "loss": 0.0008,
+      "step": 17877
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.246528703813669e-05,
+      "loss": 0.0057,
+      "step": 17878
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.245056008716909e-05,
+      "loss": 0.0076,
+      "step": 17879
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2435834765461286e-05,
+      "loss": 0.0039,
+      "step": 17880
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2421111073259245e-05,
+      "loss": 0.0018,
+      "step": 17881
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.240638901080874e-05,
+      "loss": 0.0017,
+      "step": 17882
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.239166857835587e-05,
+      "loss": 0.0009,
+      "step": 17883
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2376949776146275e-05,
+      "loss": 0.0136,
+      "step": 17884
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.236223260442601e-05,
+      "loss": 0.0054,
+      "step": 17885
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.23475170634407e-05,
+      "loss": 0.0114,
+      "step": 17886
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.23328031534362e-05,
+      "loss": 0.0199,
+      "step": 17887
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.231809087465823e-05,
+      "loss": 0.001,
+      "step": 17888
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.230338022735254e-05,
+      "loss": 0.0048,
+      "step": 17889
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2288671211764824e-05,
+      "loss": 0.0042,
+      "step": 17890
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.227396382814072e-05,
+      "loss": 0.0028,
+      "step": 17891
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.22592580767259e-05,
+      "loss": 0.0019,
+      "step": 17892
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2244553957765974e-05,
+      "loss": 0.0052,
+      "step": 17893
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.222985147150658e-05,
+      "loss": 0.0019,
+      "step": 17894
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.221515061819311e-05,
+      "loss": 0.0029,
+      "step": 17895
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2200451398071315e-05,
+      "loss": 0.0093,
+      "step": 17896
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.218575381138649e-05,
+      "loss": 0.0018,
+      "step": 17897
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2171057858384254e-05,
+      "loss": 0.0005,
+      "step": 17898
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.215636353931008e-05,
+      "loss": 0.0069,
+      "step": 17899
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.214167085440921e-05,
+      "loss": 0.0037,
+      "step": 17900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2126979803927264e-05,
+      "loss": 0.0002,
+      "step": 17901
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.2112290388109404e-05,
+      "loss": 0.005,
+      "step": 17902
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.209760260720119e-05,
+      "loss": 0.0016,
+      "step": 17903
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.208291646144775e-05,
+      "loss": 0.0041,
+      "step": 17904
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.206823195109442e-05,
+      "loss": 0.0013,
+      "step": 17905
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.20535490763865e-05,
+      "loss": 0.0019,
+      "step": 17906
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.203886783756919e-05,
+      "loss": 0.0021,
+      "step": 17907
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.20241882348877e-05,
+      "loss": 0.0041,
+      "step": 17908
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.200951026858722e-05,
+      "loss": 0.0051,
+      "step": 17909
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1994833938912884e-05,
+      "loss": 0.0049,
+      "step": 17910
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1980159246109816e-05,
+      "loss": 0.0028,
+      "step": 17911
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.196548619042311e-05,
+      "loss": 0.005,
+      "step": 17912
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.195081477209783e-05,
+      "loss": 0.0065,
+      "step": 17913
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.19361449913791e-05,
+      "loss": 0.0079,
+      "step": 17914
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.192147684851174e-05,
+      "loss": 0.0027,
+      "step": 17915
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.190681034374094e-05,
+      "loss": 0.001,
+      "step": 17916
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.189214547731154e-05,
+      "loss": 0.0035,
+      "step": 17917
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.18774822494685e-05,
+      "loss": 0.0037,
+      "step": 17918
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.186282066045673e-05,
+      "loss": 0.0019,
+      "step": 17919
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.184816071052109e-05,
+      "loss": 0.0095,
+      "step": 17920
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.183350239990644e-05,
+      "loss": 0.0001,
+      "step": 17921
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1818845728857617e-05,
+      "loss": 0.0026,
+      "step": 17922
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.180419069761939e-05,
+      "loss": 0.0135,
+      "step": 17923
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.178953730643654e-05,
+      "loss": 0.0001,
+      "step": 17924
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.17748855555538e-05,
+      "loss": 0.0015,
+      "step": 17925
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1760235445215894e-05,
+      "loss": 0.0022,
+      "step": 17926
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.174558697566755e-05,
+      "loss": 0.0034,
+      "step": 17927
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.173094014715326e-05,
+      "loss": 0.0024,
+      "step": 17928
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.171629495991781e-05,
+      "loss": 0.0043,
+      "step": 17929
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1701651414205765e-05,
+      "loss": 0.0075,
+      "step": 17930
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1687009510261687e-05,
+      "loss": 0.0023,
+      "step": 17931
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.167236924833019e-05,
+      "loss": 0.0056,
+      "step": 17932
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.165773062865562e-05,
+      "loss": 0.0017,
+      "step": 17933
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1643093651482694e-05,
+      "loss": 0.0041,
+      "step": 17934
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.162845831705567e-05,
+      "loss": 0.0063,
+      "step": 17935
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.161382462561917e-05,
+      "loss": 0.0003,
+      "step": 17936
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1599192577417465e-05,
+      "loss": 0.0148,
+      "step": 17937
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1584562172694985e-05,
+      "loss": 0.0005,
+      "step": 17938
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.156993341169611e-05,
+      "loss": 0.0091,
+      "step": 17939
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1555306294665124e-05,
+      "loss": 0.0016,
+      "step": 17940
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1540680821846366e-05,
+      "loss": 0.0025,
+      "step": 17941
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.152605699348407e-05,
+      "loss": 0.0026,
+      "step": 17942
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1511434809822535e-05,
+      "loss": 0.005,
+      "step": 17943
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1496814271105935e-05,
+      "loss": 0.0038,
+      "step": 17944
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1482195377578515e-05,
+      "loss": 0.0004,
+      "step": 17945
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.146757812948432e-05,
+      "loss": 0.0009,
+      "step": 17946
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1452962527067654e-05,
+      "loss": 0.0024,
+      "step": 17947
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.143834857057242e-05,
+      "loss": 0.0039,
+      "step": 17948
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.142373626024294e-05,
+      "loss": 0.0078,
+      "step": 17949
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1409125596323064e-05,
+      "loss": 0.0004,
+      "step": 17950
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.139451657905686e-05,
+      "loss": 0.004,
+      "step": 17951
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1379909208688464e-05,
+      "loss": 0.006,
+      "step": 17952
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.136530348546163e-05,
+      "loss": 0.0033,
+      "step": 17953
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.135069940962052e-05,
+      "loss": 0.0006,
+      "step": 17954
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1336096981408884e-05,
+      "loss": 0.0028,
+      "step": 17955
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1321496201070675e-05,
+      "loss": 0.0076,
+      "step": 17956
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.130689706884975e-05,
+      "loss": 0.0038,
+      "step": 17957
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1292299584989957e-05,
+      "loss": 0.0058,
+      "step": 17958
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.127770374973506e-05,
+      "loss": 0.0019,
+      "step": 17959
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1263109563328876e-05,
+      "loss": 0.0083,
+      "step": 17960
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1248517026015135e-05,
+      "loss": 0.005,
+      "step": 17961
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.123392613803756e-05,
+      "loss": 0.0045,
+      "step": 17962
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1219336899639864e-05,
+      "loss": 0.0024,
+      "step": 17963
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.120474931106571e-05,
+      "loss": 0.0054,
+      "step": 17964
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.119016337255877e-05,
+      "loss": 0.0016,
+      "step": 17965
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.117557908436252e-05,
+      "loss": 0.0001,
+      "step": 17966
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.116099644672074e-05,
+      "loss": 0.0043,
+      "step": 17967
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.114641545987685e-05,
+      "loss": 0.003,
+      "step": 17968
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.113183612407441e-05,
+      "loss": 0.0029,
+      "step": 17969
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1117258439556944e-05,
+      "loss": 0.0033,
+      "step": 17970
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.110268240656791e-05,
+      "loss": 0.0148,
+      "step": 17971
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1088108025350766e-05,
+      "loss": 0.001,
+      "step": 17972
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.107353529614893e-05,
+      "loss": 0.0071,
+      "step": 17973
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1058964219205786e-05,
+      "loss": 0.015,
+      "step": 17974
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.104439479476471e-05,
+      "loss": 0.0016,
+      "step": 17975
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.102982702306903e-05,
+      "loss": 0.0029,
+      "step": 17976
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.101526090436205e-05,
+      "loss": 0.0046,
+      "step": 17977
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1000696438887117e-05,
+      "loss": 0.0052,
+      "step": 17978
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.098613362688732e-05,
+      "loss": 0.0035,
+      "step": 17979
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0971572468606106e-05,
+      "loss": 0.0033,
+      "step": 17980
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.095701296428645e-05,
+      "loss": 0.0102,
+      "step": 17981
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.094245511417169e-05,
+      "loss": 0.0026,
+      "step": 17982
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0927898918504976e-05,
+      "loss": 0.0031,
+      "step": 17983
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0913344377529245e-05,
+      "loss": 0.0044,
+      "step": 17984
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.089879149148781e-05,
+      "loss": 0.0072,
+      "step": 17985
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.088424026062351e-05,
+      "loss": 0.0004,
+      "step": 17986
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.086969068517961e-05,
+      "loss": 0.0045,
+      "step": 17987
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.085514276539894e-05,
+      "loss": 0.0024,
+      "step": 17988
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0840596501524533e-05,
+      "loss": 0.0004,
+      "step": 17989
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.082605189379934e-05,
+      "loss": 0.008,
+      "step": 17990
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.081150894246629e-05,
+      "loss": 0.0043,
+      "step": 17991
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.079696764776827e-05,
+      "loss": 0.0019,
+      "step": 17992
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0782428009948145e-05,
+      "loss": 0.0016,
+      "step": 17993
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0767890029248764e-05,
+      "loss": 0.0011,
+      "step": 17994
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.075335370591293e-05,
+      "loss": 0.0086,
+      "step": 17995
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.073881904018348e-05,
+      "loss": 0.0036,
+      "step": 17996
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.072428603230304e-05,
+      "loss": 0.0006,
+      "step": 17997
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.07097546825145e-05,
+      "loss": 0.0005,
+      "step": 17998
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.069522499106037e-05,
+      "loss": 0.0062,
+      "step": 17999
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.068069695818356e-05,
+      "loss": 0.0033,
+      "step": 18000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.066617058412653e-05,
+      "loss": 0.0014,
+      "step": 18001
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0651645869131947e-05,
+      "loss": 0.0018,
+      "step": 18002
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.063712281344241e-05,
+      "loss": 0.0031,
+      "step": 18003
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.062260141730044e-05,
+      "loss": 0.0011,
+      "step": 18004
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.060808168094873e-05,
+      "loss": 0.0053,
+      "step": 18005
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.059356360462959e-05,
+      "loss": 0.0017,
+      "step": 18006
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0579047188585604e-05,
+      "loss": 0.0089,
+      "step": 18007
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.056453243305919e-05,
+      "loss": 0.0081,
+      "step": 18008
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.055001933829279e-05,
+      "loss": 0.0033,
+      "step": 18009
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0535507904528795e-05,
+      "loss": 0.0088,
+      "step": 18010
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.052099813200956e-05,
+      "loss": 0.002,
+      "step": 18011
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.050649002097743e-05,
+      "loss": 0.0023,
+      "step": 18012
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0491983571674744e-05,
+      "loss": 0.0002,
+      "step": 18013
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.047747878434375e-05,
+      "loss": 0.0009,
+      "step": 18014
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.046297565922672e-05,
+      "loss": 0.002,
+      "step": 18015
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0448474196565954e-05,
+      "loss": 0.0007,
+      "step": 18016
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.043397439660348e-05,
+      "loss": 0.005,
+      "step": 18017
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.041947625958167e-05,
+      "loss": 0.0024,
+      "step": 18018
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0404979785742536e-05,
+      "loss": 0.002,
+      "step": 18019
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0390484975328224e-05,
+      "loss": 0.002,
+      "step": 18020
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.037599182858086e-05,
+      "loss": 0.0171,
+      "step": 18021
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.036150034574247e-05,
+      "loss": 0.0029,
+      "step": 18022
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.034701052705512e-05,
+      "loss": 0.006,
+      "step": 18023
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.033252237276079e-05,
+      "loss": 0.0019,
+      "step": 18024
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.031803588310148e-05,
+      "loss": 0.0116,
+      "step": 18025
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0303551058319116e-05,
+      "loss": 0.0023,
+      "step": 18026
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.028906789865566e-05,
+      "loss": 0.0043,
+      "step": 18027
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.027458640435297e-05,
+      "loss": 0.0055,
+      "step": 18028
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.026010657565301e-05,
+      "loss": 0.0033,
+      "step": 18029
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.024562841279742e-05,
+      "loss": 0.0073,
+      "step": 18030
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.023115191602824e-05,
+      "loss": 0.0058,
+      "step": 18031
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.021667708558705e-05,
+      "loss": 0.0032,
+      "step": 18032
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.020220392171581e-05,
+      "loss": 0.0001,
+      "step": 18033
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.018773242465609e-05,
+      "loss": 0.0038,
+      "step": 18034
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.017326259464961e-05,
+      "loss": 0.0029,
+      "step": 18035
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.015879443193817e-05,
+      "loss": 0.0056,
+      "step": 18036
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.014432793676322e-05,
+      "loss": 0.0067,
+      "step": 18037
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.012986310936659e-05,
+      "loss": 0.0007,
+      "step": 18038
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.011539994998973e-05,
+      "loss": 0.0049,
+      "step": 18039
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0100938458874215e-05,
+      "loss": 0.0033,
+      "step": 18040
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.008647863626162e-05,
+      "loss": 0.0033,
+      "step": 18041
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.007202048239342e-05,
+      "loss": 0.0032,
+      "step": 18042
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.00575639975111e-05,
+      "loss": 0.0057,
+      "step": 18043
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0043109181856124e-05,
+      "loss": 0.011,
+      "step": 18044
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.00286560356699e-05,
+      "loss": 0.0017,
+      "step": 18045
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.001420455919384e-05,
+      "loss": 0.0056,
+      "step": 18046
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.999975475266936e-05,
+      "loss": 0.003,
+      "step": 18047
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.998530661633762e-05,
+      "loss": 0.0048,
+      "step": 18048
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.997086015044016e-05,
+      "loss": 0.0062,
+      "step": 18049
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.995641535521805e-05,
+      "loss": 0.0028,
+      "step": 18050
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.994197223091274e-05,
+      "loss": 0.0007,
+      "step": 18051
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.992753077776533e-05,
+      "loss": 0.0129,
+      "step": 18052
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.991309099601705e-05,
+      "loss": 0.0101,
+      "step": 18053
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.989865288590908e-05,
+      "loss": 0.0085,
+      "step": 18054
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.988421644768257e-05,
+      "loss": 0.007,
+      "step": 18055
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.986978168157861e-05,
+      "loss": 0.006,
+      "step": 18056
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.985534858783831e-05,
+      "loss": 0.0064,
+      "step": 18057
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.984091716670274e-05,
+      "loss": 0.0012,
+      "step": 18058
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.982648741841289e-05,
+      "loss": 0.0033,
+      "step": 18059
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.981205934320981e-05,
+      "loss": 0.0025,
+      "step": 18060
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9797632941334447e-05,
+      "loss": 0.0008,
+      "step": 18061
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9783208213027834e-05,
+      "loss": 0.0219,
+      "step": 18062
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.976878515853069e-05,
+      "loss": 0.0058,
+      "step": 18063
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.97543637780841e-05,
+      "loss": 0.0008,
+      "step": 18064
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.973994407192889e-05,
+      "loss": 0.0033,
+      "step": 18065
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.972552604030578e-05,
+      "loss": 0.0057,
+      "step": 18066
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.971110968345577e-05,
+      "loss": 0.0034,
+      "step": 18067
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.969669500161945e-05,
+      "loss": 0.0069,
+      "step": 18068
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9682281995037737e-05,
+      "loss": 0.0015,
+      "step": 18069
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.966787066395124e-05,
+      "loss": 0.0094,
+      "step": 18070
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9653461008600686e-05,
+      "loss": 0.0014,
+      "step": 18071
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9639053029226745e-05,
+      "loss": 0.0056,
+      "step": 18072
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.962464672607008e-05,
+      "loss": 0.0017,
+      "step": 18073
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.961024209937128e-05,
+      "loss": 0.0032,
+      "step": 18074
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.959583914937094e-05,
+      "loss": 0.0005,
+      "step": 18075
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.95814378763096e-05,
+      "loss": 0.015,
+      "step": 18076
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9567038280427807e-05,
+      "loss": 0.0018,
+      "step": 18077
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.955264036196605e-05,
+      "loss": 0.0031,
+      "step": 18078
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9538244121164824e-05,
+      "loss": 0.002,
+      "step": 18079
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.952384955826459e-05,
+      "loss": 0.0098,
+      "step": 18080
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.950945667350564e-05,
+      "loss": 0.0015,
+      "step": 18081
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9495065467128544e-05,
+      "loss": 0.0022,
+      "step": 18082
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9480675939373484e-05,
+      "loss": 0.0103,
+      "step": 18083
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.946628809048097e-05,
+      "loss": 0.0057,
+      "step": 18084
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9451901920691165e-05,
+      "loss": 0.0028,
+      "step": 18085
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.94375174302444e-05,
+      "loss": 0.0055,
+      "step": 18086
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9423134619380914e-05,
+      "loss": 0.0123,
+      "step": 18087
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.940875348834087e-05,
+      "loss": 0.0029,
+      "step": 18088
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.939437403736464e-05,
+      "loss": 0.0052,
+      "step": 18089
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.937999626669222e-05,
+      "loss": 0.0033,
+      "step": 18090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.936562017656381e-05,
+      "loss": 0.0051,
+      "step": 18091
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.935124576721947e-05,
+      "loss": 0.0037,
+      "step": 18092
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.933687303889939e-05,
+      "loss": 0.0058,
+      "step": 18093
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.932250199184343e-05,
+      "loss": 0.0015,
+      "step": 18094
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9308132626291795e-05,
+      "loss": 0.001,
+      "step": 18095
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9293764942484406e-05,
+      "loss": 0.0001,
+      "step": 18096
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.927939894066123e-05,
+      "loss": 0.0041,
+      "step": 18097
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.926503462106228e-05,
+      "loss": 0.0086,
+      "step": 18098
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.925067198392729e-05,
+      "loss": 0.01,
+      "step": 18099
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.923631102949637e-05,
+      "loss": 0.0046,
+      "step": 18100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.922195175800915e-05,
+      "loss": 0.0021,
+      "step": 18101
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.920759416970567e-05,
+      "loss": 0.0056,
+      "step": 18102
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.919323826482558e-05,
+      "loss": 0.0032,
+      "step": 18103
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.917888404360868e-05,
+      "loss": 0.0038,
+      "step": 18104
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9164531506294745e-05,
+      "loss": 0.0053,
+      "step": 18105
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.915018065312347e-05,
+      "loss": 0.0026,
+      "step": 18106
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.913583148433455e-05,
+      "loss": 0.0058,
+      "step": 18107
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.912148400016765e-05,
+      "loss": 0.0013,
+      "step": 18108
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9107138200862364e-05,
+      "loss": 0.0044,
+      "step": 18109
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9092794086658324e-05,
+      "loss": 0.0076,
+      "step": 18110
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.907845165779512e-05,
+      "loss": 0.0017,
+      "step": 18111
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.906411091451226e-05,
+      "loss": 0.0093,
+      "step": 18112
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.904977185704932e-05,
+      "loss": 0.0013,
+      "step": 18113
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.903543448564567e-05,
+      "loss": 0.0016,
+      "step": 18114
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9021098800540934e-05,
+      "loss": 0.0161,
+      "step": 18115
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.90067648019744e-05,
+      "loss": 0.0005,
+      "step": 18116
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.899243249018549e-05,
+      "loss": 0.003,
+      "step": 18117
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.897810186541374e-05,
+      "loss": 0.0008,
+      "step": 18118
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.896377292789825e-05,
+      "loss": 0.0055,
+      "step": 18119
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8949445677878596e-05,
+      "loss": 0.0023,
+      "step": 18120
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.893512011559389e-05,
+      "loss": 0.0081,
+      "step": 18121
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8920796241283444e-05,
+      "loss": 0.0086,
+      "step": 18122
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.89064740551865e-05,
+      "loss": 0.0064,
+      "step": 18123
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8892153557542284e-05,
+      "loss": 0.0083,
+      "step": 18124
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.887783474858994e-05,
+      "loss": 0.0067,
+      "step": 18125
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.886351762856864e-05,
+      "loss": 0.008,
+      "step": 18126
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8849202197717506e-05,
+      "loss": 0.0039,
+      "step": 18127
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.883488845627564e-05,
+      "loss": 0.0004,
+      "step": 18128
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.882057640448209e-05,
+      "loss": 0.0223,
+      "step": 18129
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.880626604257591e-05,
+      "loss": 0.0027,
+      "step": 18130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8791957370796146e-05,
+      "loss": 0.0007,
+      "step": 18131
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.877765038938165e-05,
+      "loss": 0.0041,
+      "step": 18132
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8763345098571546e-05,
+      "loss": 0.0086,
+      "step": 18133
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8749041498604585e-05,
+      "loss": 0.0059,
+      "step": 18134
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.873473958971987e-05,
+      "loss": 0.0071,
+      "step": 18135
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8720439372156084e-05,
+      "loss": 0.0002,
+      "step": 18136
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.870614084615214e-05,
+      "loss": 0.0007,
+      "step": 18137
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8691844011946865e-05,
+      "loss": 0.003,
+      "step": 18138
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.867754886977902e-05,
+      "loss": 0.0005,
+      "step": 18139
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.866325541988736e-05,
+      "loss": 0.0006,
+      "step": 18140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.864896366251064e-05,
+      "loss": 0.0044,
+      "step": 18141
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.863467359788753e-05,
+      "loss": 0.0047,
+      "step": 18142
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8620385226256715e-05,
+      "loss": 0.0028,
+      "step": 18143
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.860609854785687e-05,
+      "loss": 0.005,
+      "step": 18144
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.859181356292649e-05,
+      "loss": 0.0003,
+      "step": 18145
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.857753027170433e-05,
+      "loss": 0.0076,
+      "step": 18146
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8563248674428775e-05,
+      "loss": 0.0045,
+      "step": 18147
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8548968771338496e-05,
+      "loss": 0.0023,
+      "step": 18148
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.853469056267199e-05,
+      "loss": 0.0031,
+      "step": 18149
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.852041404866757e-05,
+      "loss": 0.0074,
+      "step": 18150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.85061392295639e-05,
+      "loss": 0.0036,
+      "step": 18151
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.849186610559918e-05,
+      "loss": 0.003,
+      "step": 18152
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.847759467701202e-05,
+      "loss": 0.0113,
+      "step": 18153
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.84633249440406e-05,
+      "loss": 0.0089,
+      "step": 18154
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.844905690692332e-05,
+      "loss": 0.0031,
+      "step": 18155
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8434790565898496e-05,
+      "loss": 0.0027,
+      "step": 18156
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.842052592120438e-05,
+      "loss": 0.0002,
+      "step": 18157
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8406262973079216e-05,
+      "loss": 0.0015,
+      "step": 18158
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.839200172176124e-05,
+      "loss": 0.0019,
+      "step": 18159
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8377742167488627e-05,
+      "loss": 0.0022,
+      "step": 18160
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.836348431049956e-05,
+      "loss": 0.0116,
+      "step": 18161
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8349228151032145e-05,
+      "loss": 0.0056,
+      "step": 18162
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8334973689324496e-05,
+      "loss": 0.0006,
+      "step": 18163
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8320720925614756e-05,
+      "loss": 0.0007,
+      "step": 18164
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.830646986014081e-05,
+      "loss": 0.0215,
+      "step": 18165
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.829222049314087e-05,
+      "loss": 0.0038,
+      "step": 18166
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.827797282485278e-05,
+      "loss": 0.0027,
+      "step": 18167
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8263726855514574e-05,
+      "loss": 0.0051,
+      "step": 18168
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.824948258536416e-05,
+      "loss": 0.0063,
+      "step": 18169
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.823524001463939e-05,
+      "loss": 0.0052,
+      "step": 18170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.822099914357832e-05,
+      "loss": 0.0006,
+      "step": 18171
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8206759972418654e-05,
+      "loss": 0.0032,
+      "step": 18172
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.819252250139821e-05,
+      "loss": 0.0004,
+      "step": 18173
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.817828673075485e-05,
+      "loss": 0.0274,
+      "step": 18174
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.816405266072628e-05,
+      "loss": 0.0056,
+      "step": 18175
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8149820291550264e-05,
+      "loss": 0.0002,
+      "step": 18176
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.813558962346451e-05,
+      "loss": 0.0054,
+      "step": 18177
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.81213606567067e-05,
+      "loss": 0.0064,
+      "step": 18178
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.810713339151446e-05,
+      "loss": 0.0017,
+      "step": 18179
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.809290782812543e-05,
+      "loss": 0.0062,
+      "step": 18180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8078683966777214e-05,
+      "loss": 0.0008,
+      "step": 18181
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8064461807707397e-05,
+      "loss": 0.0078,
+      "step": 18182
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8050241351153394e-05,
+      "loss": 0.0052,
+      "step": 18183
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8036022597352916e-05,
+      "loss": 0.0066,
+      "step": 18184
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.802180554654323e-05,
+      "loss": 0.0052,
+      "step": 18185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.800759019896197e-05,
+      "loss": 0.013,
+      "step": 18186
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7993376554846447e-05,
+      "loss": 0.0002,
+      "step": 18187
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7979164614434066e-05,
+      "loss": 0.0057,
+      "step": 18188
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.796495437796223e-05,
+      "loss": 0.0025,
+      "step": 18189
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7950745845668273e-05,
+      "loss": 0.004,
+      "step": 18190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.793653901778947e-05,
+      "loss": 0.0032,
+      "step": 18191
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.792233389456313e-05,
+      "loss": 0.0033,
+      "step": 18192
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.790813047622649e-05,
+      "loss": 0.0015,
+      "step": 18193
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7893928763016795e-05,
+      "loss": 0.0095,
+      "step": 18194
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.787972875517129e-05,
+      "loss": 0.0121,
+      "step": 18195
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7865530452926955e-05,
+      "loss": 0.0073,
+      "step": 18196
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.785133385652117e-05,
+      "loss": 0.0101,
+      "step": 18197
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7837138966190825e-05,
+      "loss": 0.005,
+      "step": 18198
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.78229457821732e-05,
+      "loss": 0.0067,
+      "step": 18199
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.780875430470521e-05,
+      "loss": 0.0004,
+      "step": 18200
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.779456453402387e-05,
+      "loss": 0.0096,
+      "step": 18201
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.778037647036633e-05,
+      "loss": 0.0027,
+      "step": 18202
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7766190113969357e-05,
+      "loss": 0.0023,
+      "step": 18203
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7752005465070094e-05,
+      "loss": 0.0092,
+      "step": 18204
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.773782252390528e-05,
+      "loss": 0.0071,
+      "step": 18205
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7723641290711876e-05,
+      "loss": 0.0063,
+      "step": 18206
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.770946176572672e-05,
+      "loss": 0.0033,
+      "step": 18207
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7695283949186636e-05,
+      "loss": 0.0054,
+      "step": 18208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.768110784132843e-05,
+      "loss": 0.0032,
+      "step": 18209
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.766693344238887e-05,
+      "loss": 0.0034,
+      "step": 18210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.765276075260469e-05,
+      "loss": 0.0018,
+      "step": 18211
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.763858977221258e-05,
+      "loss": 0.0049,
+      "step": 18212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.76244205014493e-05,
+      "loss": 0.0012,
+      "step": 18213
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.761025294055135e-05,
+      "loss": 0.0084,
+      "step": 18214
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.759608708975555e-05,
+      "loss": 0.0017,
+      "step": 18215
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.75819229492983e-05,
+      "loss": 0.0052,
+      "step": 18216
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7567760519416344e-05,
+      "loss": 0.0029,
+      "step": 18217
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.755359980034611e-05,
+      "loss": 0.0099,
+      "step": 18218
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.753944079232414e-05,
+      "loss": 0.0014,
+      "step": 18219
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.752528349558691e-05,
+      "loss": 0.0027,
+      "step": 18220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.751112791037088e-05,
+      "loss": 0.0091,
+      "step": 18221
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7496974036912474e-05,
+      "loss": 0.0035,
+      "step": 18222
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.748282187544811e-05,
+      "loss": 0.0037,
+      "step": 18223
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.746867142621411e-05,
+      "loss": 0.0018,
+      "step": 18224
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.745452268944685e-05,
+      "loss": 0.003,
+      "step": 18225
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.744037566538264e-05,
+      "loss": 0.0042,
+      "step": 18226
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7426230354257726e-05,
+      "loss": 0.0068,
+      "step": 18227
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7412086756308456e-05,
+      "loss": 0.0018,
+      "step": 18228
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.739794487177088e-05,
+      "loss": 0.0104,
+      "step": 18229
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7383804700881406e-05,
+      "loss": 0.0041,
+      "step": 18230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7369666243875987e-05,
+      "loss": 0.0006,
+      "step": 18231
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.735552950099093e-05,
+      "loss": 0.0056,
+      "step": 18232
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.734139447246231e-05,
+      "loss": 0.0043,
+      "step": 18233
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7327261158526116e-05,
+      "loss": 0.0029,
+      "step": 18234
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7313129559418545e-05,
+      "loss": 0.0019,
+      "step": 18235
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7298999675375466e-05,
+      "loss": 0.0128,
+      "step": 18236
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.728487150663304e-05,
+      "loss": 0.0036,
+      "step": 18237
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.727074505342711e-05,
+      "loss": 0.0079,
+      "step": 18238
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.725662031599363e-05,
+      "loss": 0.003,
+      "step": 18239
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.724249729456856e-05,
+      "loss": 0.0087,
+      "step": 18240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.722837598938773e-05,
+      "loss": 0.001,
+      "step": 18241
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.721425640068703e-05,
+      "loss": 0.0024,
+      "step": 18242
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.720013852870227e-05,
+      "loss": 0.0076,
+      "step": 18243
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7186022373669254e-05,
+      "loss": 0.0096,
+      "step": 18244
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7171907935823725e-05,
+      "loss": 0.0521,
+      "step": 18245
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.715779521540149e-05,
+      "loss": 0.0048,
+      "step": 18246
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.71436842126381e-05,
+      "loss": 0.0071,
+      "step": 18247
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.712957492776944e-05,
+      "loss": 0.0006,
+      "step": 18248
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7115467361030964e-05,
+      "loss": 0.0043,
+      "step": 18249
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.710136151265849e-05,
+      "loss": 0.0018,
+      "step": 18250
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7087257382887455e-05,
+      "loss": 0.0018,
+      "step": 18251
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.70731549719535e-05,
+      "loss": 0.0153,
+      "step": 18252
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7059054280092124e-05,
+      "loss": 0.0122,
+      "step": 18253
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.704495530753883e-05,
+      "loss": 0.0026,
+      "step": 18254
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7030858054529205e-05,
+      "loss": 0.0095,
+      "step": 18255
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.701676252129857e-05,
+      "loss": 0.0199,
+      "step": 18256
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7002668708082405e-05,
+      "loss": 0.003,
+      "step": 18257
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.698857661511609e-05,
+      "loss": 0.0016,
+      "step": 18258
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6974486242635004e-05,
+      "loss": 0.0014,
+      "step": 18259
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.696039759087446e-05,
+      "loss": 0.0059,
+      "step": 18260
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.694631066006978e-05,
+      "loss": 0.0023,
+      "step": 18261
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.693222545045622e-05,
+      "loss": 0.0036,
+      "step": 18262
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.691814196226908e-05,
+      "loss": 0.0027,
+      "step": 18263
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.690406019574359e-05,
+      "loss": 0.0007,
+      "step": 18264
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.688998015111479e-05,
+      "loss": 0.0007,
+      "step": 18265
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6875901828618064e-05,
+      "loss": 0.0017,
+      "step": 18266
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6861825228488345e-05,
+      "loss": 0.0016,
+      "step": 18267
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6847750350960924e-05,
+      "loss": 0.0115,
+      "step": 18268
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.683367719627073e-05,
+      "loss": 0.0021,
+      "step": 18269
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.681960576465288e-05,
+      "loss": 0.0061,
+      "step": 18270
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6805536056342385e-05,
+      "loss": 0.0028,
+      "step": 18271
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.679146807157421e-05,
+      "loss": 0.0008,
+      "step": 18272
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.677740181058336e-05,
+      "loss": 0.0024,
+      "step": 18273
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.676333727360472e-05,
+      "loss": 0.0091,
+      "step": 18274
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.674927446087325e-05,
+      "loss": 0.0048,
+      "step": 18275
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.673521337262378e-05,
+      "loss": 0.0023,
+      "step": 18276
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.672115400909117e-05,
+      "loss": 0.0205,
+      "step": 18277
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.670709637051025e-05,
+      "loss": 0.0019,
+      "step": 18278
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.669304045711584e-05,
+      "loss": 0.0042,
+      "step": 18279
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.667898626914258e-05,
+      "loss": 0.0003,
+      "step": 18280
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.666493380682536e-05,
+      "loss": 0.0095,
+      "step": 18281
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.665088307039871e-05,
+      "loss": 0.0049,
+      "step": 18282
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.66368340600975e-05,
+      "loss": 0.0003,
+      "step": 18283
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.662278677615621e-05,
+      "loss": 0.0018,
+      "step": 18284
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.660874121880949e-05,
+      "loss": 0.0057,
+      "step": 18285
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6594697388292054e-05,
+      "loss": 0.003,
+      "step": 18286
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6580655284838305e-05,
+      "loss": 0.0084,
+      "step": 18287
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6566614908682825e-05,
+      "loss": 0.005,
+      "step": 18288
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6552576260060135e-05,
+      "loss": 0.0074,
+      "step": 18289
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6538539339204686e-05,
+      "loss": 0.0021,
+      "step": 18290
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6524504146350924e-05,
+      "loss": 0.0002,
+      "step": 18291
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6510470681733274e-05,
+      "loss": 0.0014,
+      "step": 18292
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6496438945586106e-05,
+      "loss": 0.0073,
+      "step": 18293
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6482408938143785e-05,
+      "loss": 0.0016,
+      "step": 18294
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.646838065964064e-05,
+      "loss": 0.0032,
+      "step": 18295
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.645435411031098e-05,
+      "loss": 0.0106,
+      "step": 18296
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.644032929038908e-05,
+      "loss": 0.0053,
+      "step": 18297
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6426306200109084e-05,
+      "loss": 0.0022,
+      "step": 18298
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.64122848397054e-05,
+      "loss": 0.0031,
+      "step": 18299
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6398265209411966e-05,
+      "loss": 0.0008,
+      "step": 18300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6384247309463186e-05,
+      "loss": 0.0003,
+      "step": 18301
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6370231140093e-05,
+      "loss": 0.004,
+      "step": 18302
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6356216701535576e-05,
+      "loss": 0.0046,
+      "step": 18303
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.634220399402497e-05,
+      "loss": 0.0015,
+      "step": 18304
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.632819301779524e-05,
+      "loss": 0.01,
+      "step": 18305
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.631418377308037e-05,
+      "loss": 0.001,
+      "step": 18306
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.630017626011435e-05,
+      "loss": 0.0017,
+      "step": 18307
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.628617047913113e-05,
+      "loss": 0.0015,
+      "step": 18308
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.627216643036465e-05,
+      "loss": 0.0045,
+      "step": 18309
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6258164114048784e-05,
+      "loss": 0.0006,
+      "step": 18310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.62441635304174e-05,
+      "loss": 0.0172,
+      "step": 18311
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.623016467970439e-05,
+      "loss": 0.0064,
+      "step": 18312
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.621616756214342e-05,
+      "loss": 0.0047,
+      "step": 18313
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.62021721779684e-05,
+      "loss": 0.0008,
+      "step": 18314
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.618817852741308e-05,
+      "loss": 0.0057,
+      "step": 18315
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.617418661071105e-05,
+      "loss": 0.0006,
+      "step": 18316
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.616019642809618e-05,
+      "loss": 0.0076,
+      "step": 18317
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.614620797980193e-05,
+      "loss": 0.0056,
+      "step": 18318
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6132221266062166e-05,
+      "loss": 0.0029,
+      "step": 18319
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.611823628711031e-05,
+      "loss": 0.0203,
+      "step": 18320
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.610425304318e-05,
+      "loss": 0.0031,
+      "step": 18321
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.609027153450478e-05,
+      "loss": 0.0018,
+      "step": 18322
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.607629176131818e-05,
+      "loss": 0.0011,
+      "step": 18323
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.606231372385366e-05,
+      "loss": 0.0001,
+      "step": 18324
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.604833742234471e-05,
+      "loss": 0.0066,
+      "step": 18325
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.603436285702474e-05,
+      "loss": 0.0015,
+      "step": 18326
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.602039002812717e-05,
+      "loss": 0.0004,
+      "step": 18327
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6006418935885345e-05,
+      "loss": 0.004,
+      "step": 18328
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.599244958053264e-05,
+      "loss": 0.0065,
+      "step": 18329
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5978481962302397e-05,
+      "loss": 0.0033,
+      "step": 18330
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5964516081427765e-05,
+      "loss": 0.0094,
+      "step": 18331
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.595055193814219e-05,
+      "loss": 0.0042,
+      "step": 18332
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.593658953267871e-05,
+      "loss": 0.0059,
+      "step": 18333
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.592262886527072e-05,
+      "loss": 0.0019,
+      "step": 18334
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.590866993615123e-05,
+      "loss": 0.0054,
+      "step": 18335
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.589471274555344e-05,
+      "loss": 0.0026,
+      "step": 18336
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.588075729371046e-05,
+      "loss": 0.002,
+      "step": 18337
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5866803580855386e-05,
+      "loss": 0.0045,
+      "step": 18338
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5852851607221246e-05,
+      "loss": 0.0077,
+      "step": 18339
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.583890137304107e-05,
+      "loss": 0.0039,
+      "step": 18340
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.582495287854787e-05,
+      "loss": 0.0023,
+      "step": 18341
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.58110061239746e-05,
+      "loss": 0.0021,
+      "step": 18342
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.579706110955425e-05,
+      "loss": 0.0108,
+      "step": 18343
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.578311783551956e-05,
+      "loss": 0.0022,
+      "step": 18344
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5769176302103594e-05,
+      "loss": 0.0003,
+      "step": 18345
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5755236509539124e-05,
+      "loss": 0.0018,
+      "step": 18346
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5741298458058975e-05,
+      "loss": 0.0148,
+      "step": 18347
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5727362147896006e-05,
+      "loss": 0.0082,
+      "step": 18348
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5713427579282826e-05,
+      "loss": 0.002,
+      "step": 18349
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.569949475245234e-05,
+      "loss": 0.004,
+      "step": 18350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5685563667637075e-05,
+      "loss": 0.0084,
+      "step": 18351
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5671634325069925e-05,
+      "loss": 0.0044,
+      "step": 18352
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5657706724983344e-05,
+      "loss": 0.0026,
+      "step": 18353
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5643780867610024e-05,
+      "loss": 0.0029,
+      "step": 18354
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.562985675318256e-05,
+      "loss": 0.0017,
+      "step": 18355
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5615934381933476e-05,
+      "loss": 0.0032,
+      "step": 18356
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5602013754095344e-05,
+      "loss": 0.0024,
+      "step": 18357
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.558809486990066e-05,
+      "loss": 0.001,
+      "step": 18358
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5574177729581867e-05,
+      "loss": 0.003,
+      "step": 18359
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5560262333371434e-05,
+      "loss": 0.004,
+      "step": 18360
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5546348681501804e-05,
+      "loss": 0.0077,
+      "step": 18361
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.553243677420523e-05,
+      "loss": 0.0043,
+      "step": 18362
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.551852661171425e-05,
+      "loss": 0.0051,
+      "step": 18363
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5504618194261e-05,
+      "loss": 0.0041,
+      "step": 18364
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.549071152207797e-05,
+      "loss": 0.0002,
+      "step": 18365
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5476806595397295e-05,
+      "loss": 0.0005,
+      "step": 18366
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.54629034144512e-05,
+      "loss": 0.0039,
+      "step": 18367
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.544900197947203e-05,
+      "loss": 0.0099,
+      "step": 18368
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.543510229069181e-05,
+      "loss": 0.0018,
+      "step": 18369
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5421204348342824e-05,
+      "loss": 0.0033,
+      "step": 18370
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.540730815265711e-05,
+      "loss": 0.0136,
+      "step": 18371
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5393413703866775e-05,
+      "loss": 0.0039,
+      "step": 18372
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5379521002203884e-05,
+      "loss": 0.014,
+      "step": 18373
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.53656300479005e-05,
+      "loss": 0.0066,
+      "step": 18374
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.535174084118858e-05,
+      "loss": 0.0082,
+      "step": 18375
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.533785338230014e-05,
+      "loss": 0.0045,
+      "step": 18376
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.532396767146709e-05,
+      "loss": 0.005,
+      "step": 18377
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.53100837089214e-05,
+      "loss": 0.0028,
+      "step": 18378
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5296201494894915e-05,
+      "loss": 0.0082,
+      "step": 18379
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5282321029619505e-05,
+      "loss": 0.0027,
+      "step": 18380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.526844231332706e-05,
+      "loss": 0.0005,
+      "step": 18381
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5254565346249216e-05,
+      "loss": 0.0013,
+      "step": 18382
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.524069012861794e-05,
+      "loss": 0.0015,
+      "step": 18383
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.522681666066481e-05,
+      "loss": 0.002,
+      "step": 18384
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.521294494262171e-05,
+      "loss": 0.0043,
+      "step": 18385
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.519907497472017e-05,
+      "loss": 0.0062,
+      "step": 18386
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.51852067571919e-05,
+      "loss": 0.0179,
+      "step": 18387
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.517134029026854e-05,
+      "loss": 0.0058,
+      "step": 18388
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.515747557418167e-05,
+      "loss": 0.0104,
+      "step": 18389
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.514361260916287e-05,
+      "loss": 0.0208,
+      "step": 18390
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.512975139544366e-05,
+      "loss": 0.0107,
+      "step": 18391
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.511589193325556e-05,
+      "loss": 0.0017,
+      "step": 18392
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.510203422283006e-05,
+      "loss": 0.0032,
+      "step": 18393
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.5088178264398624e-05,
+      "loss": 0.0116,
+      "step": 18394
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.507432405819257e-05,
+      "loss": 0.0062,
+      "step": 18395
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.5060471604443444e-05,
+      "loss": 0.0027,
+      "step": 18396
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.504662090338245e-05,
+      "loss": 0.0027,
+      "step": 18397
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.503277195524104e-05,
+      "loss": 0.0031,
+      "step": 18398
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.501892476025053e-05,
+      "loss": 0.0019,
+      "step": 18399
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.500507931864204e-05,
+      "loss": 0.0089,
+      "step": 18400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4991235630647023e-05,
+      "loss": 0.0024,
+      "step": 18401
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.497739369649649e-05,
+      "loss": 0.0025,
+      "step": 18402
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.496355351642181e-05,
+      "loss": 0.0106,
+      "step": 18403
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.494971509065403e-05,
+      "loss": 0.0009,
+      "step": 18404
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.493587841942428e-05,
+      "loss": 0.0047,
+      "step": 18405
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.492204350296369e-05,
+      "loss": 0.0064,
+      "step": 18406
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4908210341503334e-05,
+      "loss": 0.0068,
+      "step": 18407
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.489437893527423e-05,
+      "loss": 0.0031,
+      "step": 18408
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.488054928450741e-05,
+      "loss": 0.0021,
+      "step": 18409
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.486672138943383e-05,
+      "loss": 0.0085,
+      "step": 18410
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.485289525028447e-05,
+      "loss": 0.0007,
+      "step": 18411
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.483907086729027e-05,
+      "loss": 0.0029,
+      "step": 18412
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.482524824068202e-05,
+      "loss": 0.0019,
+      "step": 18413
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.481142737069072e-05,
+      "loss": 0.003,
+      "step": 18414
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.479760825754707e-05,
+      "loss": 0.0034,
+      "step": 18415
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4783790901482016e-05,
+      "loss": 0.0011,
+      "step": 18416
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.476997530272623e-05,
+      "loss": 0.002,
+      "step": 18417
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.475616146151051e-05,
+      "loss": 0.0136,
+      "step": 18418
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.474234937806554e-05,
+      "loss": 0.0021,
+      "step": 18419
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.472853905262197e-05,
+      "loss": 0.0049,
+      "step": 18420
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4714730485410624e-05,
+      "loss": 0.0032,
+      "step": 18421
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.470092367666196e-05,
+      "loss": 0.0213,
+      "step": 18422
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.468711862660662e-05,
+      "loss": 0.0002,
+      "step": 18423
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4673315335475204e-05,
+      "loss": 0.0071,
+      "step": 18424
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.465951380349825e-05,
+      "loss": 0.0019,
+      "step": 18425
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.464571403090626e-05,
+      "loss": 0.0054,
+      "step": 18426
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.463191601792971e-05,
+      "loss": 0.0039,
+      "step": 18427
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.461811976479906e-05,
+      "loss": 0.0001,
+      "step": 18428
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.460432527174473e-05,
+      "loss": 0.0027,
+      "step": 18429
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.459053253899712e-05,
+      "loss": 0.0021,
+      "step": 18430
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.457674156678658e-05,
+      "loss": 0.0041,
+      "step": 18431
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.456295235534352e-05,
+      "loss": 0.0075,
+      "step": 18432
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.454916490489807e-05,
+      "loss": 0.0049,
+      "step": 18433
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.453537921568073e-05,
+      "loss": 0.0011,
+      "step": 18434
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4521595287921566e-05,
+      "loss": 0.0057,
+      "step": 18435
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.450781312185089e-05,
+      "loss": 0.0064,
+      "step": 18436
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4494032717698855e-05,
+      "loss": 0.0021,
+      "step": 18437
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4480254075695624e-05,
+      "loss": 0.0103,
+      "step": 18438
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4466477196071345e-05,
+      "loss": 0.0032,
+      "step": 18439
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.44527020790561e-05,
+      "loss": 0.0038,
+      "step": 18440
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.443892872487997e-05,
+      "loss": 0.0018,
+      "step": 18441
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4425157133772976e-05,
+      "loss": 0.0028,
+      "step": 18442
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.441138730596516e-05,
+      "loss": 0.0046,
+      "step": 18443
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4397619241686474e-05,
+      "loss": 0.0024,
+      "step": 18444
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4383852941166953e-05,
+      "loss": 0.0071,
+      "step": 18445
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.437008840463635e-05,
+      "loss": 0.0074,
+      "step": 18446
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4356325632324755e-05,
+      "loss": 0.0006,
+      "step": 18447
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4342564624461826e-05,
+      "loss": 0.0058,
+      "step": 18448
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.432880538127763e-05,
+      "loss": 0.0038,
+      "step": 18449
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4315047903001795e-05,
+      "loss": 0.0117,
+      "step": 18450
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.430129218986411e-05,
+      "loss": 0.0063,
+      "step": 18451
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4287538242094454e-05,
+      "loss": 0.0023,
+      "step": 18452
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4273786059922354e-05,
+      "loss": 0.0104,
+      "step": 18453
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.426003564357771e-05,
+      "loss": 0.0033,
+      "step": 18454
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.424628699329002e-05,
+      "loss": 0.0017,
+      "step": 18455
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.423254010928896e-05,
+      "loss": 0.0116,
+      "step": 18456
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.421879499180413e-05,
+      "loss": 0.0043,
+      "step": 18457
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4205051641065096e-05,
+      "loss": 0.0056,
+      "step": 18458
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.41913100573014e-05,
+      "loss": 0.0013,
+      "step": 18459
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4177570240742566e-05,
+      "loss": 0.0006,
+      "step": 18460
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.416383219161805e-05,
+      "loss": 0.0041,
+      "step": 18461
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4150095910157314e-05,
+      "loss": 0.0059,
+      "step": 18462
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4136361396589845e-05,
+      "loss": 0.0008,
+      "step": 18463
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.412262865114487e-05,
+      "loss": 0.0062,
+      "step": 18464
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4108897674051945e-05,
+      "loss": 0.0028,
+      "step": 18465
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.409516846554022e-05,
+      "loss": 0.0016,
+      "step": 18466
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4081441025839194e-05,
+      "loss": 0.0024,
+      "step": 18467
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.406771535517798e-05,
+      "loss": 0.0001,
+      "step": 18468
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.405399145378588e-05,
+      "loss": 0.001,
+      "step": 18469
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.404026932189211e-05,
+      "loss": 0.0031,
+      "step": 18470
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.402654895972588e-05,
+      "loss": 0.0054,
+      "step": 18471
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.401283036751629e-05,
+      "loss": 0.0044,
+      "step": 18472
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.399911354549252e-05,
+      "loss": 0.0006,
+      "step": 18473
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.398539849388362e-05,
+      "loss": 0.0089,
+      "step": 18474
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.397168521291871e-05,
+      "loss": 0.0043,
+      "step": 18475
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.395797370282678e-05,
+      "loss": 0.0006,
+      "step": 18476
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.394426396383687e-05,
+      "loss": 0.0088,
+      "step": 18477
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3930555996177986e-05,
+      "loss": 0.0017,
+      "step": 18478
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3916849800078936e-05,
+      "loss": 0.0057,
+      "step": 18479
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.39031453757688e-05,
+      "loss": 0.0044,
+      "step": 18480
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3889442723476395e-05,
+      "loss": 0.0036,
+      "step": 18481
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3875741843430614e-05,
+      "loss": 0.0048,
+      "step": 18482
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.386204273586029e-05,
+      "loss": 0.0014,
+      "step": 18483
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.384834540099414e-05,
+      "loss": 0.0042,
+      "step": 18484
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.383464983906107e-05,
+      "loss": 0.0011,
+      "step": 18485
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.38209560502897e-05,
+      "loss": 0.004,
+      "step": 18486
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.38072640349088e-05,
+      "loss": 0.0004,
+      "step": 18487
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3793573793147035e-05,
+      "loss": 0.0171,
+      "step": 18488
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377988532523307e-05,
+      "loss": 0.0007,
+      "step": 18489
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3766198631395526e-05,
+      "loss": 0.0022,
+      "step": 18490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3752513711863e-05,
+      "loss": 0.001,
+      "step": 18491
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3738830566864044e-05,
+      "loss": 0.0008,
+      "step": 18492
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.37251491966272e-05,
+      "loss": 0.0029,
+      "step": 18493
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3711469601380984e-05,
+      "loss": 0.0077,
+      "step": 18494
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.369779178135384e-05,
+      "loss": 0.0063,
+      "step": 18495
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.368411573677429e-05,
+      "loss": 0.0013,
+      "step": 18496
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.367044146787061e-05,
+      "loss": 0.0041,
+      "step": 18497
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.365676897487135e-05,
+      "loss": 0.0027,
+      "step": 18498
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.364309825800469e-05,
+      "loss": 0.0024,
+      "step": 18499
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3629429317499145e-05,
+      "loss": 0.0029,
+      "step": 18500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.361576215358288e-05,
+      "loss": 0.001,
+      "step": 18501
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3602096766484196e-05,
+      "loss": 0.005,
+      "step": 18502
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3588433156431333e-05,
+      "loss": 0.0082,
+      "step": 18503
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.357477132365247e-05,
+      "loss": 0.0071,
+      "step": 18504
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.35611112683759e-05,
+      "loss": 0.0086,
+      "step": 18505
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.354745299082964e-05,
+      "loss": 0.0115,
+      "step": 18506
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.353379649124186e-05,
+      "loss": 0.0094,
+      "step": 18507
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.352014176984065e-05,
+      "loss": 0.0015,
+      "step": 18508
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.350648882685412e-05,
+      "loss": 0.0103,
+      "step": 18509
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.349283766251015e-05,
+      "loss": 0.0048,
+      "step": 18510
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.347918827703687e-05,
+      "loss": 0.0053,
+      "step": 18511
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3465540670662245e-05,
+      "loss": 0.0003,
+      "step": 18512
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.345189484361417e-05,
+      "loss": 0.0024,
+      "step": 18513
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.343825079612063e-05,
+      "loss": 0.0034,
+      "step": 18514
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.342460852840935e-05,
+      "loss": 0.0091,
+      "step": 18515
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.341096804070839e-05,
+      "loss": 0.0033,
+      "step": 18516
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.339732933324535e-05,
+      "loss": 0.0061,
+      "step": 18517
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.338369240624824e-05,
+      "loss": 0.0011,
+      "step": 18518
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.337005725994468e-05,
+      "loss": 0.0076,
+      "step": 18519
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.335642389456244e-05,
+      "loss": 0.0035,
+      "step": 18520
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.334279231032921e-05,
+      "loss": 0.0002,
+      "step": 18521
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.332916250747269e-05,
+      "loss": 0.005,
+      "step": 18522
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.331553448622052e-05,
+      "loss": 0.0017,
+      "step": 18523
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3301908246800296e-05,
+      "loss": 0.0032,
+      "step": 18524
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.328828378943961e-05,
+      "loss": 0.0038,
+      "step": 18525
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3274661114366024e-05,
+      "loss": 0.0026,
+      "step": 18526
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.326104022180705e-05,
+      "loss": 0.0011,
+      "step": 18527
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.324742111199019e-05,
+      "loss": 0.0012,
+      "step": 18528
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.323380378514295e-05,
+      "loss": 0.0059,
+      "step": 18529
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.322018824149264e-05,
+      "loss": 0.0037,
+      "step": 18530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.320657448126683e-05,
+      "loss": 0.0069,
+      "step": 18531
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.319296250469271e-05,
+      "loss": 0.0064,
+      "step": 18532
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.317935231199781e-05,
+      "loss": 0.0002,
+      "step": 18533
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3165743903409377e-05,
+      "loss": 0.0022,
+      "step": 18534
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.315213727915461e-05,
+      "loss": 0.0037,
+      "step": 18535
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3138532439460915e-05,
+      "loss": 0.0005,
+      "step": 18536
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3124929384555415e-05,
+      "loss": 0.0031,
+      "step": 18537
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.311132811466534e-05,
+      "loss": 0.0105,
+      "step": 18538
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.309772863001783e-05,
+      "loss": 0.002,
+      "step": 18539
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.308413093084006e-05,
+      "loss": 0.0009,
+      "step": 18540
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3070535017359116e-05,
+      "loss": 0.0059,
+      "step": 18541
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.305694088980208e-05,
+      "loss": 0.001,
+      "step": 18542
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3043348548396e-05,
+      "loss": 0.0049,
+      "step": 18543
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3029757993367906e-05,
+      "loss": 0.0038,
+      "step": 18544
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.301616922494477e-05,
+      "loss": 0.013,
+      "step": 18545
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.300258224335355e-05,
+      "loss": 0.0038,
+      "step": 18546
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2988997048821214e-05,
+      "loss": 0.0065,
+      "step": 18547
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.297541364157455e-05,
+      "loss": 0.0009,
+      "step": 18548
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2961832021840575e-05,
+      "loss": 0.0045,
+      "step": 18549
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2948252189845974e-05,
+      "loss": 0.0051,
+      "step": 18550
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.293467414581773e-05,
+      "loss": 0.0047,
+      "step": 18551
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2921097889982474e-05,
+      "loss": 0.0034,
+      "step": 18552
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.290752342256702e-05,
+      "loss": 0.0052,
+      "step": 18553
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.289395074379808e-05,
+      "loss": 0.0092,
+      "step": 18554
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2880379853902316e-05,
+      "loss": 0.0223,
+      "step": 18555
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.286681075310643e-05,
+      "loss": 0.009,
+      "step": 18556
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2853243441637025e-05,
+      "loss": 0.0113,
+      "step": 18557
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.283967791972071e-05,
+      "loss": 0.0092,
+      "step": 18558
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.282611418758406e-05,
+      "loss": 0.0207,
+      "step": 18559
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.281255224545364e-05,
+      "loss": 0.0047,
+      "step": 18560
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.279899209355586e-05,
+      "loss": 0.0034,
+      "step": 18561
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.278543373211734e-05,
+      "loss": 0.0003,
+      "step": 18562
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2771877161364364e-05,
+      "loss": 0.0033,
+      "step": 18563
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2758322381523494e-05,
+      "loss": 0.004,
+      "step": 18564
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.274476939282113e-05,
+      "loss": 0.002,
+      "step": 18565
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.273121819548349e-05,
+      "loss": 0.0091,
+      "step": 18566
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2717668789737055e-05,
+      "loss": 0.0045,
+      "step": 18567
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2704121175807984e-05,
+      "loss": 0.0029,
+      "step": 18568
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.269057535392273e-05,
+      "loss": 0.0033,
+      "step": 18569
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.267703132430737e-05,
+      "loss": 0.0073,
+      "step": 18570
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.266348908718818e-05,
+      "loss": 0.0085,
+      "step": 18571
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2649948642791326e-05,
+      "loss": 0.0018,
+      "step": 18572
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.263640999134296e-05,
+      "loss": 0.001,
+      "step": 18573
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.262287313306922e-05,
+      "loss": 0.0049,
+      "step": 18574
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.26093380681962e-05,
+      "loss": 0.0039,
+      "step": 18575
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2595804796949916e-05,
+      "loss": 0.005,
+      "step": 18576
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2582273319556454e-05,
+      "loss": 0.0045,
+      "step": 18577
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2568743636241794e-05,
+      "loss": 0.0004,
+      "step": 18578
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.25552157472319e-05,
+      "loss": 0.0054,
+      "step": 18579
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.254168965275277e-05,
+      "loss": 0.0133,
+      "step": 18580
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.252816535303016e-05,
+      "loss": 0.0019,
+      "step": 18581
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.251464284829017e-05,
+      "loss": 0.0104,
+      "step": 18582
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2501122138758477e-05,
+      "loss": 0.0089,
+      "step": 18583
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2487603224660965e-05,
+      "loss": 0.0037,
+      "step": 18584
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.247408610622343e-05,
+      "loss": 0.004,
+      "step": 18585
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.246057078367158e-05,
+      "loss": 0.0042,
+      "step": 18586
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.244705725723127e-05,
+      "loss": 0.0051,
+      "step": 18587
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2433545527128094e-05,
+      "loss": 0.0072,
+      "step": 18588
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2420035593587746e-05,
+      "loss": 0.0016,
+      "step": 18589
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.240652745683587e-05,
+      "loss": 0.0008,
+      "step": 18590
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.23930211170981e-05,
+      "loss": 0.0028,
+      "step": 18591
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.237951657459998e-05,
+      "loss": 0.0003,
+      "step": 18592
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.236601382956709e-05,
+      "loss": 0.008,
+      "step": 18593
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.235251288222495e-05,
+      "loss": 0.002,
+      "step": 18594
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.233901373279904e-05,
+      "loss": 0.0033,
+      "step": 18595
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.232551638151481e-05,
+      "loss": 0.0026,
+      "step": 18596
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.231202082859773e-05,
+      "loss": 0.0031,
+      "step": 18597
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.22985270742732e-05,
+      "loss": 0.0022,
+      "step": 18598
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2285035118766496e-05,
+      "loss": 0.0025,
+      "step": 18599
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.227154496230311e-05,
+      "loss": 0.0031,
+      "step": 18600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2258056605108195e-05,
+      "loss": 0.0005,
+      "step": 18601
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.22445700474072e-05,
+      "loss": 0.0033,
+      "step": 18602
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.223108528942524e-05,
+      "loss": 0.0071,
+      "step": 18603
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.221760233138757e-05,
+      "loss": 0.0056,
+      "step": 18604
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2204121173519424e-05,
+      "loss": 0.0013,
+      "step": 18605
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.219064181604591e-05,
+      "loss": 0.0092,
+      "step": 18606
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.217716425919219e-05,
+      "loss": 0.006,
+      "step": 18607
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.216368850318337e-05,
+      "loss": 0.005,
+      "step": 18608
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.215021454824449e-05,
+      "loss": 0.0025,
+      "step": 18609
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2136742394600606e-05,
+      "loss": 0.0057,
+      "step": 18610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2123272042476785e-05,
+      "loss": 0.0008,
+      "step": 18611
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.210980349209787e-05,
+      "loss": 0.0025,
+      "step": 18612
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.209633674368896e-05,
+      "loss": 0.0035,
+      "step": 18613
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.208287179747485e-05,
+      "loss": 0.002,
+      "step": 18614
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.206940865368055e-05,
+      "loss": 0.0027,
+      "step": 18615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.205594731253083e-05,
+      "loss": 0.0053,
+      "step": 18616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2042487774250497e-05,
+      "loss": 0.0092,
+      "step": 18617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.202903003906449e-05,
+      "loss": 0.0059,
+      "step": 18618
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2015574107197414e-05,
+      "loss": 0.0024,
+      "step": 18619
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.200211997887417e-05,
+      "loss": 0.0072,
+      "step": 18620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1988667654319335e-05,
+      "loss": 0.0135,
+      "step": 18621
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1975217133757626e-05,
+      "loss": 0.0077,
+      "step": 18622
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1961768417413724e-05,
+      "loss": 0.0112,
+      "step": 18623
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.194832150551221e-05,
+      "loss": 0.0062,
+      "step": 18624
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.19348763982777e-05,
+      "loss": 0.0012,
+      "step": 18625
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.192143309593473e-05,
+      "loss": 0.0029,
+      "step": 18626
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.190799159870785e-05,
+      "loss": 0.0047,
+      "step": 18627
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.189455190682153e-05,
+      "loss": 0.0094,
+      "step": 18628
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.188111402050028e-05,
+      "loss": 0.0034,
+      "step": 18629
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.186767793996851e-05,
+      "loss": 0.0085,
+      "step": 18630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.185424366545066e-05,
+      "loss": 0.0028,
+      "step": 18631
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1840811197171006e-05,
+      "loss": 0.0073,
+      "step": 18632
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.182738053535406e-05,
+      "loss": 0.0031,
+      "step": 18633
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1813951680224e-05,
+      "loss": 0.0006,
+      "step": 18634
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.180052463200516e-05,
+      "loss": 0.0033,
+      "step": 18635
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1787099390921815e-05,
+      "loss": 0.0032,
+      "step": 18636
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.177367595719816e-05,
+      "loss": 0.0015,
+      "step": 18637
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1760254331058404e-05,
+      "loss": 0.0052,
+      "step": 18638
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.174683451272674e-05,
+      "loss": 0.0041,
+      "step": 18639
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.173341650242726e-05,
+      "loss": 0.0006,
+      "step": 18640
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.17200003003841e-05,
+      "loss": 0.0019,
+      "step": 18641
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1706585906821334e-05,
+      "loss": 0.0077,
+      "step": 18642
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1693173321962985e-05,
+      "loss": 0.0018,
+      "step": 18643
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1679762546033145e-05,
+      "loss": 0.0034,
+      "step": 18644
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1666353579255654e-05,
+      "loss": 0.0096,
+      "step": 18645
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.165294642185459e-05,
+      "loss": 0.0015,
+      "step": 18646
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1639541074053826e-05,
+      "loss": 0.0033,
+      "step": 18647
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.162613753607729e-05,
+      "loss": 0.0054,
+      "step": 18648
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.161273580814887e-05,
+      "loss": 0.0043,
+      "step": 18649
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.159933589049226e-05,
+      "loss": 0.0048,
+      "step": 18650
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.158593778333147e-05,
+      "loss": 0.0067,
+      "step": 18651
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.157254148689006e-05,
+      "loss": 0.0005,
+      "step": 18652
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.155914700139197e-05,
+      "loss": 0.0002,
+      "step": 18653
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1545754327060765e-05,
+      "loss": 0.0165,
+      "step": 18654
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.15323634641202e-05,
+      "loss": 0.0029,
+      "step": 18655
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1518974412793906e-05,
+      "loss": 0.0073,
+      "step": 18656
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.150558717330552e-05,
+      "loss": 0.0017,
+      "step": 18657
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1492201745878604e-05,
+      "loss": 0.0032,
+      "step": 18658
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1478818130736767e-05,
+      "loss": 0.0227,
+      "step": 18659
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.146543632810351e-05,
+      "loss": 0.0062,
+      "step": 18660
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.145205633820233e-05,
+      "loss": 0.0091,
+      "step": 18661
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1438678161256764e-05,
+      "loss": 0.0018,
+      "step": 18662
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.142530179749009e-05,
+      "loss": 0.0075,
+      "step": 18663
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.141192724712594e-05,
+      "loss": 0.0029,
+      "step": 18664
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1398554510387476e-05,
+      "loss": 0.0003,
+      "step": 18665
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1385183587498235e-05,
+      "loss": 0.0009,
+      "step": 18666
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1371814478681406e-05,
+      "loss": 0.0019,
+      "step": 18667
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.135844718416034e-05,
+      "loss": 0.0022,
+      "step": 18668
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1345081704158275e-05,
+      "loss": 0.0011,
+      "step": 18669
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.13317180388984e-05,
+      "loss": 0.0055,
+      "step": 18670
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1318356188604056e-05,
+      "loss": 0.0068,
+      "step": 18671
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.130499615349826e-05,
+      "loss": 0.0064,
+      "step": 18672
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.129163793380422e-05,
+      "loss": 0.0064,
+      "step": 18673
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.127828152974502e-05,
+      "loss": 0.0023,
+      "step": 18674
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.126492694154373e-05,
+      "loss": 0.0051,
+      "step": 18675
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.125157416942343e-05,
+      "loss": 0.0044,
+      "step": 18676
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.123822321360712e-05,
+      "loss": 0.0033,
+      "step": 18677
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.122487407431777e-05,
+      "loss": 0.0016,
+      "step": 18678
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1211526751778354e-05,
+      "loss": 0.0004,
+      "step": 18679
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.119818124621181e-05,
+      "loss": 0.0031,
+      "step": 18680
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1184837557841e-05,
+      "loss": 0.0049,
+      "step": 18681
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.117149568688887e-05,
+      "loss": 0.0022,
+      "step": 18682
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.115815563357809e-05,
+      "loss": 0.0006,
+      "step": 18683
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1144817398131634e-05,
+      "loss": 0.0076,
+      "step": 18684
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.113148098077218e-05,
+      "loss": 0.0024,
+      "step": 18685
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1118146381722514e-05,
+      "loss": 0.0015,
+      "step": 18686
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.11048136012053e-05,
+      "loss": 0.0126,
+      "step": 18687
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.1091482639443266e-05,
+      "loss": 0.0002,
+      "step": 18688
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.107815349665906e-05,
+      "loss": 0.0146,
+      "step": 18689
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.1064826173075284e-05,
+      "loss": 0.0037,
+      "step": 18690
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.1051500668914554e-05,
+      "loss": 0.0048,
+      "step": 18691
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.1038176984399415e-05,
+      "loss": 0.0006,
+      "step": 18692
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.102485511975239e-05,
+      "loss": 0.0097,
+      "step": 18693
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.101153507519598e-05,
+      "loss": 0.0023,
+      "step": 18694
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0998216850952745e-05,
+      "loss": 0.003,
+      "step": 18695
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0984900447244924e-05,
+      "loss": 0.0032,
+      "step": 18696
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0971585864295145e-05,
+      "loss": 0.0026,
+      "step": 18697
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.095827310232559e-05,
+      "loss": 0.0093,
+      "step": 18698
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.094496216155877e-05,
+      "loss": 0.0054,
+      "step": 18699
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.093165304221696e-05,
+      "loss": 0.0079,
+      "step": 18700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.091834574452235e-05,
+      "loss": 0.0049,
+      "step": 18701
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.090504026869736e-05,
+      "loss": 0.0164,
+      "step": 18702
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.089173661496404e-05,
+      "loss": 0.004,
+      "step": 18703
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.087843478354476e-05,
+      "loss": 0.0006,
+      "step": 18704
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0865134774661553e-05,
+      "loss": 0.0028,
+      "step": 18705
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.085183658853661e-05,
+      "loss": 0.0044,
+      "step": 18706
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.083854022539202e-05,
+      "loss": 0.0051,
+      "step": 18707
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0825245685449866e-05,
+      "loss": 0.0007,
+      "step": 18708
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.081195296893221e-05,
+      "loss": 0.0002,
+      "step": 18709
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.079866207606101e-05,
+      "loss": 0.0037,
+      "step": 18710
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.078537300705831e-05,
+      "loss": 0.0011,
+      "step": 18711
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.077208576214604e-05,
+      "loss": 0.0061,
+      "step": 18712
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.075880034154616e-05,
+      "loss": 0.0066,
+      "step": 18713
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.074551674548043e-05,
+      "loss": 0.0106,
+      "step": 18714
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0732234974170894e-05,
+      "loss": 0.0073,
+      "step": 18715
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0718955027839206e-05,
+      "loss": 0.0057,
+      "step": 18716
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0705676906707326e-05,
+      "loss": 0.0021,
+      "step": 18717
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.069240061099692e-05,
+      "loss": 0.0006,
+      "step": 18718
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0679126140929754e-05,
+      "loss": 0.0107,
+      "step": 18719
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.066585349672754e-05,
+      "loss": 0.0037,
+      "step": 18720
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0652582678611954e-05,
+      "loss": 0.0078,
+      "step": 18721
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.063931368680464e-05,
+      "loss": 0.0031,
+      "step": 18722
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0626046521527225e-05,
+      "loss": 0.0067,
+      "step": 18723
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0612781183001285e-05,
+      "loss": 0.0042,
+      "step": 18724
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.059951767144839e-05,
+      "loss": 0.0108,
+      "step": 18725
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.058625598709005e-05,
+      "loss": 0.0039,
+      "step": 18726
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.057299613014777e-05,
+      "loss": 0.0029,
+      "step": 18727
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.055973810084306e-05,
+      "loss": 0.0033,
+      "step": 18728
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.054648189939722e-05,
+      "loss": 0.0001,
+      "step": 18729
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.053322752603178e-05,
+      "loss": 0.0041,
+      "step": 18730
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.051997498096808e-05,
+      "loss": 0.004,
+      "step": 18731
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.050672426442745e-05,
+      "loss": 0.0015,
+      "step": 18732
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.049347537663126e-05,
+      "loss": 0.0022,
+      "step": 18733
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.048022831780065e-05,
+      "loss": 0.0124,
+      "step": 18734
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.046698308815704e-05,
+      "loss": 0.0037,
+      "step": 18735
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.045373968792154e-05,
+      "loss": 0.0018,
+      "step": 18736
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.044049811731535e-05,
+      "loss": 0.0041,
+      "step": 18737
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.042725837655967e-05,
+      "loss": 0.0061,
+      "step": 18738
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.041402046587559e-05,
+      "loss": 0.0014,
+      "step": 18739
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0400784385484244e-05,
+      "loss": 0.0048,
+      "step": 18740
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.038755013560668e-05,
+      "loss": 0.0013,
+      "step": 18741
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.037431771646393e-05,
+      "loss": 0.006,
+      "step": 18742
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0361087128277e-05,
+      "loss": 0.0113,
+      "step": 18743
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.034785837126688e-05,
+      "loss": 0.0029,
+      "step": 18744
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.033463144565451e-05,
+      "loss": 0.0007,
+      "step": 18745
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0321406351660846e-05,
+      "loss": 0.0039,
+      "step": 18746
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0308183089506644e-05,
+      "loss": 0.0065,
+      "step": 18747
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.029496165941293e-05,
+      "loss": 0.0035,
+      "step": 18748
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.028174206160035e-05,
+      "loss": 0.0004,
+      "step": 18749
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.026852429628986e-05,
+      "loss": 0.0017,
+      "step": 18750
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.025530836370211e-05,
+      "loss": 0.0038,
+      "step": 18751
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0242094264057835e-05,
+      "loss": 0.0012,
+      "step": 18752
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.022888199757785e-05,
+      "loss": 0.0055,
+      "step": 18753
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0215671564482654e-05,
+      "loss": 0.0012,
+      "step": 18754
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.020246296499307e-05,
+      "loss": 0.0061,
+      "step": 18755
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0189256199329564e-05,
+      "loss": 0.0003,
+      "step": 18756
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.017605126771277e-05,
+      "loss": 0.0094,
+      "step": 18757
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.016284817036321e-05,
+      "loss": 0.0061,
+      "step": 18758
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.014964690750148e-05,
+      "loss": 0.0082,
+      "step": 18759
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0136447479347904e-05,
+      "loss": 0.0018,
+      "step": 18760
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.012324988612309e-05,
+      "loss": 0.017,
+      "step": 18761
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.01100541280474e-05,
+      "loss": 0.0079,
+      "step": 18762
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.009686020534124e-05,
+      "loss": 0.0129,
+      "step": 18763
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.008366811822501e-05,
+      "loss": 0.011,
+      "step": 18764
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.007047786691893e-05,
+      "loss": 0.0064,
+      "step": 18765
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0057289451643466e-05,
+      "loss": 0.0003,
+      "step": 18766
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.00441028726187e-05,
+      "loss": 0.0052,
+      "step": 18767
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.003091813006506e-05,
+      "loss": 0.0007,
+      "step": 18768
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.001773522420263e-05,
+      "loss": 0.0034,
+      "step": 18769
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.000455415525162e-05,
+      "loss": 0.0061,
+      "step": 18770
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9991374923432195e-05,
+      "loss": 0.0056,
+      "step": 18771
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.997819752896446e-05,
+      "loss": 0.0005,
+      "step": 18772
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.996502197206852e-05,
+      "loss": 0.0067,
+      "step": 18773
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.99518482529644e-05,
+      "loss": 0.0043,
+      "step": 18774
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9938676371872155e-05,
+      "loss": 0.0045,
+      "step": 18775
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9925506329011767e-05,
+      "loss": 0.0042,
+      "step": 18776
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.991233812460321e-05,
+      "loss": 0.005,
+      "step": 18777
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9899171758866406e-05,
+      "loss": 0.0018,
+      "step": 18778
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.988600723202131e-05,
+      "loss": 0.0025,
+      "step": 18779
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9872844544287654e-05,
+      "loss": 0.0032,
+      "step": 18780
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.985968369588547e-05,
+      "loss": 0.0029,
+      "step": 18781
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.984652468703443e-05,
+      "loss": 0.0002,
+      "step": 18782
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.983336751795431e-05,
+      "loss": 0.0074,
+      "step": 18783
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9820212188864984e-05,
+      "loss": 0.0117,
+      "step": 18784
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9807058699986014e-05,
+      "loss": 0.0043,
+      "step": 18785
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9793907051537263e-05,
+      "loss": 0.0048,
+      "step": 18786
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.978075724373824e-05,
+      "loss": 0.0074,
+      "step": 18787
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.976760927680863e-05,
+      "loss": 0.0017,
+      "step": 18788
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9754463150968016e-05,
+      "loss": 0.0081,
+      "step": 18789
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.974131886643598e-05,
+      "loss": 0.0072,
+      "step": 18790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.972817642343203e-05,
+      "loss": 0.0036,
+      "step": 18791
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.971503582217571e-05,
+      "loss": 0.0004,
+      "step": 18792
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.970189706288645e-05,
+      "loss": 0.0001,
+      "step": 18793
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9688760145783724e-05,
+      "loss": 0.0079,
+      "step": 18794
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9675625071086934e-05,
+      "loss": 0.0109,
+      "step": 18795
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.966249183901544e-05,
+      "loss": 0.0027,
+      "step": 18796
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.964936044978867e-05,
+      "loss": 0.0021,
+      "step": 18797
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.963623090362579e-05,
+      "loss": 0.0042,
+      "step": 18798
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9623103200746265e-05,
+      "loss": 0.0004,
+      "step": 18799
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.960997734136918e-05,
+      "loss": 0.002,
+      "step": 18800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.959685332571393e-05,
+      "loss": 0.0008,
+      "step": 18801
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.95837311539996e-05,
+      "loss": 0.0016,
+      "step": 18802
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9570610826445375e-05,
+      "loss": 0.0088,
+      "step": 18803
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.95574923432704e-05,
+      "loss": 0.004,
+      "step": 18804
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.954437570469379e-05,
+      "loss": 0.0042,
+      "step": 18805
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9531260910934586e-05,
+      "loss": 0.0023,
+      "step": 18806
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.951814796221186e-05,
+      "loss": 0.0031,
+      "step": 18807
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9505036858744615e-05,
+      "loss": 0.0036,
+      "step": 18808
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9491927600751824e-05,
+      "loss": 0.0068,
+      "step": 18809
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9478820188452485e-05,
+      "loss": 0.0044,
+      "step": 18810
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.946571462206539e-05,
+      "loss": 0.0071,
+      "step": 18811
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.945261090180958e-05,
+      "loss": 0.002,
+      "step": 18812
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.943950902790377e-05,
+      "loss": 0.0003,
+      "step": 18813
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9426409000566906e-05,
+      "loss": 0.0021,
+      "step": 18814
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9413310820017765e-05,
+      "loss": 0.0024,
+      "step": 18815
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9400214486474983e-05,
+      "loss": 0.0037,
+      "step": 18816
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9387120000157496e-05,
+      "loss": 0.0033,
+      "step": 18817
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9374027361283825e-05,
+      "loss": 0.0025,
+      "step": 18818
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.936093657007279e-05,
+      "loss": 0.0013,
+      "step": 18819
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.93478476267429e-05,
+      "loss": 0.0021,
+      "step": 18820
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9334760531512855e-05,
+      "loss": 0.0075,
+      "step": 18821
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.932167528460119e-05,
+      "loss": 0.0016,
+      "step": 18822
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.930859188622646e-05,
+      "loss": 0.0061,
+      "step": 18823
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9295510336607204e-05,
+      "loss": 0.005,
+      "step": 18824
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.92824306359619e-05,
+      "loss": 0.0085,
+      "step": 18825
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.926935278450898e-05,
+      "loss": 0.0053,
+      "step": 18826
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9256276782466905e-05,
+      "loss": 0.0048,
+      "step": 18827
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.924320263005404e-05,
+      "loss": 0.0049,
+      "step": 18828
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.923013032748876e-05,
+      "loss": 0.0023,
+      "step": 18829
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9217059874989455e-05,
+      "loss": 0.0051,
+      "step": 18830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.920399127277426e-05,
+      "loss": 0.0091,
+      "step": 18831
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.919092452106165e-05,
+      "loss": 0.0161,
+      "step": 18832
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.917785962006972e-05,
+      "loss": 0.0017,
+      "step": 18833
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9164796570016735e-05,
+      "loss": 0.0013,
+      "step": 18834
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9151735371120854e-05,
+      "loss": 0.0301,
+      "step": 18835
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.913867602360019e-05,
+      "loss": 0.0017,
+      "step": 18836
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9125618527673e-05,
+      "loss": 0.0002,
+      "step": 18837
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9112562883557216e-05,
+      "loss": 0.0021,
+      "step": 18838
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9099509091470944e-05,
+      "loss": 0.0037,
+      "step": 18839
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.908645715163223e-05,
+      "loss": 0.0003,
+      "step": 18840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9073407064259026e-05,
+      "loss": 0.0021,
+      "step": 18841
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.906035882956932e-05,
+      "loss": 0.0038,
+      "step": 18842
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.904731244778102e-05,
+      "loss": 0.0041,
+      "step": 18843
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.9034267919112066e-05,
+      "loss": 0.0023,
+      "step": 18844
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.902122524378028e-05,
+      "loss": 0.0091,
+      "step": 18845
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.900818442200354e-05,
+      "loss": 0.0021,
+      "step": 18846
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.899514545399961e-05,
+      "loss": 0.0017,
+      "step": 18847
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.898210833998634e-05,
+      "loss": 0.0018,
+      "step": 18848
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.896907308018134e-05,
+      "loss": 0.0057,
+      "step": 18849
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.895603967480248e-05,
+      "loss": 0.005,
+      "step": 18850
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.894300812406729e-05,
+      "loss": 0.0058,
+      "step": 18851
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.892997842819358e-05,
+      "loss": 0.0006,
+      "step": 18852
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.891695058739886e-05,
+      "loss": 0.0084,
+      "step": 18853
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8903924601900736e-05,
+      "loss": 0.0082,
+      "step": 18854
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8890900471916774e-05,
+      "loss": 0.0028,
+      "step": 18855
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.887787819766453e-05,
+      "loss": 0.002,
+      "step": 18856
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.886485777936146e-05,
+      "loss": 0.0038,
+      "step": 18857
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.885183921722505e-05,
+      "loss": 0.0015,
+      "step": 18858
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.883882251147273e-05,
+      "loss": 0.001,
+      "step": 18859
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8825807662321925e-05,
+      "loss": 0.0005,
+      "step": 18860
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.881279466999001e-05,
+      "loss": 0.0048,
+      "step": 18861
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8799783534694225e-05,
+      "loss": 0.0023,
+      "step": 18862
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8786774256652055e-05,
+      "loss": 0.001,
+      "step": 18863
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.877376683608058e-05,
+      "loss": 0.0009,
+      "step": 18864
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.876076127319725e-05,
+      "loss": 0.006,
+      "step": 18865
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.874775756821915e-05,
+      "loss": 0.0042,
+      "step": 18866
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8734755721363444e-05,
+      "loss": 0.0062,
+      "step": 18867
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.872175573284744e-05,
+      "loss": 0.0005,
+      "step": 18868
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8708757602888074e-05,
+      "loss": 0.0014,
+      "step": 18869
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.869576133170263e-05,
+      "loss": 0.008,
+      "step": 18870
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.868276691950803e-05,
+      "loss": 0.0072,
+      "step": 18871
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.866977436652135e-05,
+      "loss": 0.0049,
+      "step": 18872
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8656783672959585e-05,
+      "loss": 0.0029,
+      "step": 18873
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.864379483903971e-05,
+      "loss": 0.0019,
+      "step": 18874
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.863080786497865e-05,
+      "loss": 0.0066,
+      "step": 18875
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8617822750993324e-05,
+      "loss": 0.003,
+      "step": 18876
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8604839497300626e-05,
+      "loss": 0.0079,
+      "step": 18877
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.859185810411736e-05,
+      "loss": 0.0082,
+      "step": 18878
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.857887857166037e-05,
+      "loss": 0.0039,
+      "step": 18879
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.856590090014643e-05,
+      "loss": 0.0079,
+      "step": 18880
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8552925089792346e-05,
+      "loss": 0.004,
+      "step": 18881
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.853995114081469e-05,
+      "loss": 0.0039,
+      "step": 18882
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.852697905343032e-05,
+      "loss": 0.0026,
+      "step": 18883
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.851400882785579e-05,
+      "loss": 0.0004,
+      "step": 18884
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.850104046430775e-05,
+      "loss": 0.0026,
+      "step": 18885
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.848807396300281e-05,
+      "loss": 0.01,
+      "step": 18886
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8475109324157515e-05,
+      "loss": 0.0035,
+      "step": 18887
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.846214654798843e-05,
+      "loss": 0.0016,
+      "step": 18888
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.844918563471204e-05,
+      "loss": 0.01,
+      "step": 18889
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.84362265845448e-05,
+      "loss": 0.0024,
+      "step": 18890
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8423269397703175e-05,
+      "loss": 0.0017,
+      "step": 18891
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.841031407440358e-05,
+      "loss": 0.0017,
+      "step": 18892
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8397360614862365e-05,
+      "loss": 0.0041,
+      "step": 18893
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.838440901929594e-05,
+      "loss": 0.0027,
+      "step": 18894
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.837145928792048e-05,
+      "loss": 0.0053,
+      "step": 18895
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.835851142095242e-05,
+      "loss": 0.007,
+      "step": 18896
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.834556541860796e-05,
+      "loss": 0.0048,
+      "step": 18897
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.833262128110332e-05,
+      "loss": 0.0061,
+      "step": 18898
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.831967900865472e-05,
+      "loss": 0.0003,
+      "step": 18899
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.830673860147821e-05,
+      "loss": 0.0122,
+      "step": 18900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.82938000597901e-05,
+      "loss": 0.0009,
+      "step": 18901
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8280863383806295e-05,
+      "loss": 0.0159,
+      "step": 18902
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.826792857374304e-05,
+      "loss": 0.005,
+      "step": 18903
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.825499562981623e-05,
+      "loss": 0.0046,
+      "step": 18904
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.824206455224195e-05,
+      "loss": 0.0028,
+      "step": 18905
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8229135341236146e-05,
+      "loss": 0.0014,
+      "step": 18906
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8216207997014755e-05,
+      "loss": 0.0239,
+      "step": 18907
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.82032825197937e-05,
+      "loss": 0.0074,
+      "step": 18908
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.819035890978886e-05,
+      "loss": 0.0051,
+      "step": 18909
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8177437167216094e-05,
+      "loss": 0.0049,
+      "step": 18910
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8164517292291185e-05,
+      "loss": 0.0023,
+      "step": 18911
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.815159928523e-05,
+      "loss": 0.0043,
+      "step": 18912
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.813868314624815e-05,
+      "loss": 0.0065,
+      "step": 18913
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.812576887556153e-05,
+      "loss": 0.006,
+      "step": 18914
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.811285647338566e-05,
+      "loss": 0.0049,
+      "step": 18915
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.809994593993637e-05,
+      "loss": 0.0006,
+      "step": 18916
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.808703727542917e-05,
+      "loss": 0.0025,
+      "step": 18917
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.80741304800797e-05,
+      "loss": 0.004,
+      "step": 18918
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.806122555410352e-05,
+      "loss": 0.0128,
+      "step": 18919
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.804832249771612e-05,
+      "loss": 0.0077,
+      "step": 18920
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.803542131113317e-05,
+      "loss": 0.0016,
+      "step": 18921
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8022521994569974e-05,
+      "loss": 0.0027,
+      "step": 18922
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.800962454824202e-05,
+      "loss": 0.0028,
+      "step": 18923
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.799672897236475e-05,
+      "loss": 0.0087,
+      "step": 18924
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.798383526715354e-05,
+      "loss": 0.0059,
+      "step": 18925
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.797094343282371e-05,
+      "loss": 0.0095,
+      "step": 18926
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.795805346959059e-05,
+      "loss": 0.0041,
+      "step": 18927
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7945165377669475e-05,
+      "loss": 0.0125,
+      "step": 18928
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7932279157275626e-05,
+      "loss": 0.0041,
+      "step": 18929
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7919394808624284e-05,
+      "loss": 0.0237,
+      "step": 18930
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7906512331930555e-05,
+      "loss": 0.0024,
+      "step": 18931
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.789363172740972e-05,
+      "loss": 0.0065,
+      "step": 18932
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7880752995276785e-05,
+      "loss": 0.0041,
+      "step": 18933
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.786787613574699e-05,
+      "loss": 0.0005,
+      "step": 18934
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.785500114903529e-05,
+      "loss": 0.0039,
+      "step": 18935
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.784212803535675e-05,
+      "loss": 0.0071,
+      "step": 18936
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7829256794926394e-05,
+      "loss": 0.0005,
+      "step": 18937
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.78163874279592e-05,
+      "loss": 0.0082,
+      "step": 18938
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.780351993467007e-05,
+      "loss": 0.0078,
+      "step": 18939
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.779065431527396e-05,
+      "loss": 0.0013,
+      "step": 18940
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7777790569985725e-05,
+      "loss": 0.0102,
+      "step": 18941
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.776492869902023e-05,
+      "loss": 0.0005,
+      "step": 18942
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.775206870259228e-05,
+      "loss": 0.0144,
+      "step": 18943
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.773921058091667e-05,
+      "loss": 0.0024,
+      "step": 18944
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7726354334208185e-05,
+      "loss": 0.0014,
+      "step": 18945
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.771349996268143e-05,
+      "loss": 0.0037,
+      "step": 18946
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.770064746655128e-05,
+      "loss": 0.0038,
+      "step": 18947
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.76877968460322e-05,
+      "loss": 0.0071,
+      "step": 18948
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.767494810133896e-05,
+      "loss": 0.0048,
+      "step": 18949
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.766210123268616e-05,
+      "loss": 0.0056,
+      "step": 18950
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.764925624028823e-05,
+      "loss": 0.0028,
+      "step": 18951
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.763641312435989e-05,
+      "loss": 0.0013,
+      "step": 18952
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7623571885115465e-05,
+      "loss": 0.0075,
+      "step": 18953
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.761073252276961e-05,
+      "loss": 0.0015,
+      "step": 18954
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.759789503753662e-05,
+      "loss": 0.0023,
+      "step": 18955
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7585059429630955e-05,
+      "loss": 0.0022,
+      "step": 18956
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7572225699266993e-05,
+      "loss": 0.0012,
+      "step": 18957
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7559393846659086e-05,
+      "loss": 0.0017,
+      "step": 18958
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7546563872021554e-05,
+      "loss": 0.0032,
+      "step": 18959
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.753373577556867e-05,
+      "loss": 0.0017,
+      "step": 18960
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.75209095575147e-05,
+      "loss": 0.0072,
+      "step": 18961
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7508085218073863e-05,
+      "loss": 0.0034,
+      "step": 18962
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.749526275746039e-05,
+      "loss": 0.0027,
+      "step": 18963
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.74824421758883e-05,
+      "loss": 0.0029,
+      "step": 18964
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7469623473571914e-05,
+      "loss": 0.0022,
+      "step": 18965
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.745680665072513e-05,
+      "loss": 0.0018,
+      "step": 18966
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.744399170756221e-05,
+      "loss": 0.001,
+      "step": 18967
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.743117864429705e-05,
+      "loss": 0.0021,
+      "step": 18968
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.741836746114371e-05,
+      "loss": 0.0042,
+      "step": 18969
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.740555815831612e-05,
+      "loss": 0.0002,
+      "step": 18970
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7392750736028265e-05,
+      "loss": 0.0029,
+      "step": 18971
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7379945194494026e-05,
+      "loss": 0.0023,
+      "step": 18972
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.736714153392729e-05,
+      "loss": 0.0082,
+      "step": 18973
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7354339754541905e-05,
+      "loss": 0.0032,
+      "step": 18974
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.734153985655167e-05,
+      "loss": 0.0053,
+      "step": 18975
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.732874184017038e-05,
+      "loss": 0.0032,
+      "step": 18976
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.731594570561179e-05,
+      "loss": 0.0066,
+      "step": 18977
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7303151453089645e-05,
+      "loss": 0.0054,
+      "step": 18978
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7290359082817514e-05,
+      "loss": 0.0019,
+      "step": 18979
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.727756859500919e-05,
+      "loss": 0.0015,
+      "step": 18980
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.72647799898783e-05,
+      "loss": 0.0072,
+      "step": 18981
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7251993267638304e-05,
+      "loss": 0.0008,
+      "step": 18982
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7239208428502925e-05,
+      "loss": 0.0077,
+      "step": 18983
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.722642547268552e-05,
+      "loss": 0.0036,
+      "step": 18984
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.721364440039979e-05,
+      "loss": 0.0023,
+      "step": 18985
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7200865211859036e-05,
+      "loss": 0.0057,
+      "step": 18986
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.718808790727676e-05,
+      "loss": 0.0031,
+      "step": 18987
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.717531248686637e-05,
+      "loss": 0.0005,
+      "step": 18988
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.716253895084124e-05,
+      "loss": 0.0036,
+      "step": 18989
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7149767299414686e-05,
+      "loss": 0.0008,
+      "step": 18990
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7136997532800053e-05,
+      "loss": 0.0016,
+      "step": 18991
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.71242296512106e-05,
+      "loss": 0.0036,
+      "step": 18992
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.711146365485959e-05,
+      "loss": 0.0024,
+      "step": 18993
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.709869954396022e-05,
+      "loss": 0.004,
+      "step": 18994
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7085937318725675e-05,
+      "loss": 0.008,
+      "step": 18995
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.707317697936917e-05,
+      "loss": 0.0022,
+      "step": 18996
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.706041852610369e-05,
+      "loss": 0.011,
+      "step": 18997
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7047661959142495e-05,
+      "loss": 0.0024,
+      "step": 18998
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7034907278698464e-05,
+      "loss": 0.0018,
+      "step": 18999
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7022154484984815e-05,
+      "loss": 0.0202,
+      "step": 19000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.700940357821439e-05,
+      "loss": 0.0033,
+      "step": 19001
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.699665455860018e-05,
+      "loss": 0.0009,
+      "step": 19002
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.698390742635523e-05,
+      "loss": 0.0067,
+      "step": 19003
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.697116218169232e-05,
+      "loss": 0.0107,
+      "step": 19004
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.695841882482435e-05,
+      "loss": 0.0015,
+      "step": 19005
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.694567735596418e-05,
+      "loss": 0.0029,
+      "step": 19006
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.693293777532458e-05,
+      "loss": 0.0033,
+      "step": 19007
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.692020008311837e-05,
+      "loss": 0.0042,
+      "step": 19008
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.690746427955826e-05,
+      "loss": 0.0004,
+      "step": 19009
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.689473036485698e-05,
+      "loss": 0.0044,
+      "step": 19010
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.688199833922719e-05,
+      "loss": 0.006,
+      "step": 19011
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.686926820288158e-05,
+      "loss": 0.0052,
+      "step": 19012
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.685653995603273e-05,
+      "loss": 0.0044,
+      "step": 19013
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.684381359889326e-05,
+      "loss": 0.0096,
+      "step": 19014
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.683108913167563e-05,
+      "loss": 0.0076,
+      "step": 19015
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.681836655459252e-05,
+      "loss": 0.0018,
+      "step": 19016
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.680564586785625e-05,
+      "loss": 0.0045,
+      "step": 19017
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6792927071679443e-05,
+      "loss": 0.0002,
+      "step": 19018
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.678021016627441e-05,
+      "loss": 0.0024,
+      "step": 19019
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6767495151853576e-05,
+      "loss": 0.0057,
+      "step": 19020
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.675478202862933e-05,
+      "loss": 0.0027,
+      "step": 19021
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.674207079681397e-05,
+      "loss": 0.002,
+      "step": 19022
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.672936145661983e-05,
+      "loss": 0.0034,
+      "step": 19023
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6716654008259155e-05,
+      "loss": 0.0045,
+      "step": 19024
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.670394845194422e-05,
+      "loss": 0.0119,
+      "step": 19025
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6691244787887183e-05,
+      "loss": 0.003,
+      "step": 19026
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.667854301630025e-05,
+      "loss": 0.002,
+      "step": 19027
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6665843137395556e-05,
+      "loss": 0.0062,
+      "step": 19028
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.665314515138527e-05,
+      "loss": 0.0013,
+      "step": 19029
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6640449058481326e-05,
+      "loss": 0.0058,
+      "step": 19030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6627754858895945e-05,
+      "loss": 0.0015,
+      "step": 19031
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.661506255284102e-05,
+      "loss": 0.0031,
+      "step": 19032
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.660237214052855e-05,
+      "loss": 0.0012,
+      "step": 19033
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6589683622170596e-05,
+      "loss": 0.0011,
+      "step": 19034
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6576996997978916e-05,
+      "loss": 0.0072,
+      "step": 19035
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.656431226816557e-05,
+      "loss": 0.0029,
+      "step": 19036
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6551629432942294e-05,
+      "loss": 0.0038,
+      "step": 19037
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.653894849252095e-05,
+      "loss": 0.0015,
+      "step": 19038
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.652626944711333e-05,
+      "loss": 0.0004,
+      "step": 19039
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.651359229693123e-05,
+      "loss": 0.008,
+      "step": 19040
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.650091704218634e-05,
+      "loss": 0.0021,
+      "step": 19041
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.64882436830904e-05,
+      "loss": 0.0025,
+      "step": 19042
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.647557221985504e-05,
+      "loss": 0.0081,
+      "step": 19043
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6462902652691935e-05,
+      "loss": 0.0006,
+      "step": 19044
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.645023498181267e-05,
+      "loss": 0.0089,
+      "step": 19045
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.64375692074288e-05,
+      "loss": 0.0063,
+      "step": 19046
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.642490532975196e-05,
+      "loss": 0.0079,
+      "step": 19047
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6412243348993486e-05,
+      "loss": 0.0032,
+      "step": 19048
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.639958326536505e-05,
+      "loss": 0.0024,
+      "step": 19049
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6386925079077916e-05,
+      "loss": 0.0099,
+      "step": 19050
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.637426879034368e-05,
+      "loss": 0.0063,
+      "step": 19051
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.636161439937361e-05,
+      "loss": 0.0055,
+      "step": 19052
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.634896190637908e-05,
+      "loss": 0.0021,
+      "step": 19053
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.63363113115714e-05,
+      "loss": 0.0088,
+      "step": 19054
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.63236626151619e-05,
+      "loss": 0.0057,
+      "step": 19055
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.63110158173618e-05,
+      "loss": 0.0047,
+      "step": 19056
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.629837091838236e-05,
+      "loss": 0.0073,
+      "step": 19057
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6285727918434726e-05,
+      "loss": 0.0042,
+      "step": 19058
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6273086817730105e-05,
+      "loss": 0.0102,
+      "step": 19059
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.626044761647965e-05,
+      "loss": 0.009,
+      "step": 19060
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.624781031489433e-05,
+      "loss": 0.002,
+      "step": 19061
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.623517491318534e-05,
+      "loss": 0.0026,
+      "step": 19062
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6222541411563666e-05,
+      "loss": 0.007,
+      "step": 19063
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.620990981024035e-05,
+      "loss": 0.008,
+      "step": 19064
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.619728010942634e-05,
+      "loss": 0.0045,
+      "step": 19065
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6184652309332503e-05,
+      "loss": 0.0011,
+      "step": 19066
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.617202641016991e-05,
+      "loss": 0.0133,
+      "step": 19067
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.615940241214924e-05,
+      "loss": 0.0027,
+      "step": 19068
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.614678031548153e-05,
+      "loss": 0.0015,
+      "step": 19069
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6134160120377464e-05,
+      "loss": 0.0044,
+      "step": 19070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.612154182704785e-05,
+      "loss": 0.0066,
+      "step": 19071
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.610892543570346e-05,
+      "loss": 0.0039,
+      "step": 19072
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6096310946555e-05,
+      "loss": 0.0072,
+      "step": 19073
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6083698359813134e-05,
+      "loss": 0.0059,
+      "step": 19074
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6071087675688555e-05,
+      "loss": 0.002,
+      "step": 19075
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.605847889439187e-05,
+      "loss": 0.0025,
+      "step": 19076
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.604587201613366e-05,
+      "loss": 0.009,
+      "step": 19077
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.603326704112454e-05,
+      "loss": 0.0008,
+      "step": 19078
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6020663969574895e-05,
+      "loss": 0.0016,
+      "step": 19079
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.600806280169541e-05,
+      "loss": 0.0041,
+      "step": 19080
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5995463537696356e-05,
+      "loss": 0.0029,
+      "step": 19081
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.598286617778836e-05,
+      "loss": 0.001,
+      "step": 19082
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5970270722181654e-05,
+      "loss": 0.0005,
+      "step": 19083
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.595767717108668e-05,
+      "loss": 0.002,
+      "step": 19084
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.594508552471378e-05,
+      "loss": 0.0017,
+      "step": 19085
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.59324957832732e-05,
+      "loss": 0.0057,
+      "step": 19086
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5919907946975334e-05,
+      "loss": 0.0044,
+      "step": 19087
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.590732201603031e-05,
+      "loss": 0.0018,
+      "step": 19088
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5894737990648374e-05,
+      "loss": 0.0051,
+      "step": 19089
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5882155871039694e-05,
+      "loss": 0.0017,
+      "step": 19090
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.586957565741444e-05,
+      "loss": 0.0008,
+      "step": 19091
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.58569973499827e-05,
+      "loss": 0.0087,
+      "step": 19092
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5844420948954565e-05,
+      "loss": 0.0012,
+      "step": 19093
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.583184645454008e-05,
+      "loss": 0.003,
+      "step": 19094
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.581927386694928e-05,
+      "loss": 0.0013,
+      "step": 19095
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.580670318639212e-05,
+      "loss": 0.0115,
+      "step": 19096
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.579413441307858e-05,
+      "loss": 0.0037,
+      "step": 19097
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5781567547218615e-05,
+      "loss": 0.0018,
+      "step": 19098
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5769002589022e-05,
+      "loss": 0.0046,
+      "step": 19099
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.575643953869875e-05,
+      "loss": 0.0083,
+      "step": 19100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5743878396458516e-05,
+      "loss": 0.0013,
+      "step": 19101
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.573131916251127e-05,
+      "loss": 0.0044,
+      "step": 19102
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5718761837066664e-05,
+      "loss": 0.0015,
+      "step": 19103
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5706206420334445e-05,
+      "loss": 0.0018,
+      "step": 19104
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.569365291252431e-05,
+      "loss": 0.001,
+      "step": 19105
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5681101313845964e-05,
+      "loss": 0.0005,
+      "step": 19106
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.566855162450901e-05,
+      "loss": 0.0024,
+      "step": 19107
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.565600384472306e-05,
+      "loss": 0.0069,
+      "step": 19108
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5643457974697686e-05,
+      "loss": 0.0052,
+      "step": 19109
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5630914014642424e-05,
+      "loss": 0.0051,
+      "step": 19110
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5618371964766825e-05,
+      "loss": 0.0019,
+      "step": 19111
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.560583182528025e-05,
+      "loss": 0.0011,
+      "step": 19112
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.559329359639229e-05,
+      "loss": 0.0028,
+      "step": 19113
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.55807572783122e-05,
+      "loss": 0.0009,
+      "step": 19114
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.55682228712495e-05,
+      "loss": 0.0123,
+      "step": 19115
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5555690375413515e-05,
+      "loss": 0.0086,
+      "step": 19116
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.554315979101345e-05,
+      "loss": 0.0022,
+      "step": 19117
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.553063111825875e-05,
+      "loss": 0.0136,
+      "step": 19118
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5518104357358503e-05,
+      "loss": 0.0044,
+      "step": 19119
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.55055795085221e-05,
+      "loss": 0.0039,
+      "step": 19120
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5493056571958605e-05,
+      "loss": 0.0041,
+      "step": 19121
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.548053554787722e-05,
+      "loss": 0.0184,
+      "step": 19122
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5468016436487045e-05,
+      "loss": 0.0036,
+      "step": 19123
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.54554992379972e-05,
+      "loss": 0.0068,
+      "step": 19124
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.544298395261672e-05,
+      "loss": 0.0089,
+      "step": 19125
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5430470580554665e-05,
+      "loss": 0.0081,
+      "step": 19126
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.541795912202001e-05,
+      "loss": 0.003,
+      "step": 19127
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.540544957722173e-05,
+      "loss": 0.0056,
+      "step": 19128
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5392941946368794e-05,
+      "loss": 0.0014,
+      "step": 19129
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5380436229669985e-05,
+      "loss": 0.0041,
+      "step": 19130
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.536793242733432e-05,
+      "loss": 0.0009,
+      "step": 19131
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5355430539570496e-05,
+      "loss": 0.0014,
+      "step": 19132
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5342930566587466e-05,
+      "loss": 0.0006,
+      "step": 19133
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.533043250859389e-05,
+      "loss": 0.003,
+      "step": 19134
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.531793636579855e-05,
+      "loss": 0.0053,
+      "step": 19135
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.530544213841014e-05,
+      "loss": 0.007,
+      "step": 19136
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.529294982663736e-05,
+      "loss": 0.0143,
+      "step": 19137
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5280459430688833e-05,
+      "loss": 0.0033,
+      "step": 19138
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.526797095077319e-05,
+      "loss": 0.0007,
+      "step": 19139
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5255484387098994e-05,
+      "loss": 0.0023,
+      "step": 19140
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5242999739874814e-05,
+      "loss": 0.0044,
+      "step": 19141
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5230517009309144e-05,
+      "loss": 0.0041,
+      "step": 19142
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5218036195610495e-05,
+      "loss": 0.003,
+      "step": 19143
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.520555729898734e-05,
+      "loss": 0.0022,
+      "step": 19144
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.519308031964799e-05,
+      "loss": 0.0087,
+      "step": 19145
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.518060525780095e-05,
+      "loss": 0.0001,
+      "step": 19146
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.516813211365454e-05,
+      "loss": 0.002,
+      "step": 19147
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.515566088741708e-05,
+      "loss": 0.0026,
+      "step": 19148
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.514319157929692e-05,
+      "loss": 0.0026,
+      "step": 19149
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.513072418950219e-05,
+      "loss": 0.0119,
+      "step": 19150
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.511825871824126e-05,
+      "loss": 0.0088,
+      "step": 19151
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5105795165722245e-05,
+      "loss": 0.0082,
+      "step": 19152
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.509333353215331e-05,
+      "loss": 0.0015,
+      "step": 19153
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.508087381774261e-05,
+      "loss": 0.0012,
+      "step": 19154
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.506841602269824e-05,
+      "loss": 0.0042,
+      "step": 19155
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.505596014722829e-05,
+      "loss": 0.0006,
+      "step": 19156
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5043506191540754e-05,
+      "loss": 0.005,
+      "step": 19157
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5031054155843686e-05,
+      "loss": 0.0009,
+      "step": 19158
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5018604040345014e-05,
+      "loss": 0.009,
+      "step": 19159
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5006155845252695e-05,
+      "loss": 0.0043,
+      "step": 19160
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.499370957077465e-05,
+      "loss": 0.0033,
+      "step": 19161
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4981265217118784e-05,
+      "loss": 0.0001,
+      "step": 19162
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4968822784492815e-05,
+      "loss": 0.0053,
+      "step": 19163
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.495638227310474e-05,
+      "loss": 0.0033,
+      "step": 19164
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.494394368316213e-05,
+      "loss": 0.0053,
+      "step": 19165
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.493150701487296e-05,
+      "loss": 0.0062,
+      "step": 19166
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.491907226844477e-05,
+      "loss": 0.0092,
+      "step": 19167
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4906639444085255e-05,
+      "loss": 0.0101,
+      "step": 19168
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.489420854200221e-05,
+      "loss": 0.0061,
+      "step": 19169
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4881779562403075e-05,
+      "loss": 0.0008,
+      "step": 19170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4869352505495606e-05,
+      "loss": 0.0012,
+      "step": 19171
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.485692737148723e-05,
+      "loss": 0.0034,
+      "step": 19172
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.484450416058551e-05,
+      "loss": 0.0026,
+      "step": 19173
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4832082872997955e-05,
+      "loss": 0.0033,
+      "step": 19174
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4819663508932e-05,
+      "loss": 0.0008,
+      "step": 19175
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.480724606859509e-05,
+      "loss": 0.0019,
+      "step": 19176
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4794830552194605e-05,
+      "loss": 0.0046,
+      "step": 19177
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.478241695993794e-05,
+      "loss": 0.0033,
+      "step": 19178
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.477000529203238e-05,
+      "loss": 0.0037,
+      "step": 19179
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.475759554868528e-05,
+      "loss": 0.0074,
+      "step": 19180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.474518773010378e-05,
+      "loss": 0.0187,
+      "step": 19181
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.473278183649531e-05,
+      "loss": 0.0015,
+      "step": 19182
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4720377868066874e-05,
+      "loss": 0.0009,
+      "step": 19183
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4707975825025803e-05,
+      "loss": 0.0012,
+      "step": 19184
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.469557570757913e-05,
+      "loss": 0.0114,
+      "step": 19185
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.468317751593401e-05,
+      "loss": 0.0024,
+      "step": 19186
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.467078125029748e-05,
+      "loss": 0.0002,
+      "step": 19187
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.465838691087662e-05,
+      "loss": 0.0055,
+      "step": 19188
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.464599449787841e-05,
+      "loss": 0.0045,
+      "step": 19189
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.463360401150984e-05,
+      "loss": 0.0006,
+      "step": 19190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4621215451977866e-05,
+      "loss": 0.0058,
+      "step": 19191
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.460882881948938e-05,
+      "loss": 0.0098,
+      "step": 19192
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4596444114251267e-05,
+      "loss": 0.0011,
+      "step": 19193
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.458406133647037e-05,
+      "loss": 0.0005,
+      "step": 19194
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.457168048635356e-05,
+      "loss": 0.0002,
+      "step": 19195
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4559301564107486e-05,
+      "loss": 0.0048,
+      "step": 19196
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.454692456993907e-05,
+      "loss": 0.0143,
+      "step": 19197
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.453454950405486e-05,
+      "loss": 0.0162,
+      "step": 19198
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.452217636666168e-05,
+      "loss": 0.002,
+      "step": 19199
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.450980515796617e-05,
+      "loss": 0.0067,
+      "step": 19200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.449743587817483e-05,
+      "loss": 0.0014,
+      "step": 19201
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.448506852749442e-05,
+      "loss": 0.0015,
+      "step": 19202
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.447270310613138e-05,
+      "loss": 0.0023,
+      "step": 19203
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.446033961429227e-05,
+      "loss": 0.0004,
+      "step": 19204
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.444797805218359e-05,
+      "loss": 0.0051,
+      "step": 19205
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.44356184200118e-05,
+      "loss": 0.0028,
+      "step": 19206
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.442326071798331e-05,
+      "loss": 0.0052,
+      "step": 19207
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.441090494630455e-05,
+      "loss": 0.0056,
+      "step": 19208
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4398551105181847e-05,
+      "loss": 0.0017,
+      "step": 19209
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4386199194821566e-05,
+      "loss": 0.0088,
+      "step": 19210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.437384921542999e-05,
+      "loss": 0.0012,
+      "step": 19211
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.436150116721339e-05,
+      "loss": 0.0002,
+      "step": 19212
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.434915505037806e-05,
+      "loss": 0.0011,
+      "step": 19213
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.433681086513006e-05,
+      "loss": 0.0048,
+      "step": 19214
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.432446861167572e-05,
+      "loss": 0.0011,
+      "step": 19215
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.431212829022102e-05,
+      "loss": 0.0053,
+      "step": 19216
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.429978990097226e-05,
+      "loss": 0.0109,
+      "step": 19217
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4287453444135354e-05,
+      "loss": 0.009,
+      "step": 19218
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.42751189199164e-05,
+      "loss": 0.0043,
+      "step": 19219
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.42627863285214e-05,
+      "loss": 0.0077,
+      "step": 19220
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.42504556701563e-05,
+      "loss": 0.0016,
+      "step": 19221
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.423812694502717e-05,
+      "loss": 0.0076,
+      "step": 19222
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.422580015333979e-05,
+      "loss": 0.0061,
+      "step": 19223
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.421347529530008e-05,
+      "loss": 0.0002,
+      "step": 19224
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.420115237111391e-05,
+      "loss": 0.0032,
+      "step": 19225
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.418883138098712e-05,
+      "loss": 0.0041,
+      "step": 19226
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4176512325125364e-05,
+      "loss": 0.0066,
+      "step": 19227
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4164195203734565e-05,
+      "loss": 0.001,
+      "step": 19228
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.415188001702029e-05,
+      "loss": 0.0026,
+      "step": 19229
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.413956676518832e-05,
+      "loss": 0.0022,
+      "step": 19230
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4127255448444346e-05,
+      "loss": 0.0016,
+      "step": 19231
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.411494606699386e-05,
+      "loss": 0.0148,
+      "step": 19232
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.410263862104258e-05,
+      "loss": 0.0064,
+      "step": 19233
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.409033311079594e-05,
+      "loss": 0.0026,
+      "step": 19234
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4078029536459604e-05,
+      "loss": 0.0041,
+      "step": 19235
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.406572789823896e-05,
+      "loss": 0.0035,
+      "step": 19236
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.40534281963395e-05,
+      "loss": 0.0041,
+      "step": 19237
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.404113043096665e-05,
+      "loss": 0.003,
+      "step": 19238
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4028834602325815e-05,
+      "loss": 0.002,
+      "step": 19239
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.401654071062235e-05,
+      "loss": 0.0021,
+      "step": 19240
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.400424875606159e-05,
+      "loss": 0.0035,
+      "step": 19241
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3991958738848836e-05,
+      "loss": 0.0059,
+      "step": 19242
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3979670659189345e-05,
+      "loss": 0.0008,
+      "step": 19243
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.396738451728836e-05,
+      "loss": 0.0041,
+      "step": 19244
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.395510031335108e-05,
+      "loss": 0.0071,
+      "step": 19245
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3942818047582725e-05,
+      "loss": 0.0004,
+      "step": 19246
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.39305377201883e-05,
+      "loss": 0.002,
+      "step": 19247
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3918259331373056e-05,
+      "loss": 0.0084,
+      "step": 19248
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.390598288134194e-05,
+      "loss": 0.0024,
+      "step": 19249
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3893708370300117e-05,
+      "loss": 0.0024,
+      "step": 19250
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.38814357984525e-05,
+      "loss": 0.017,
+      "step": 19251
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3869165166004046e-05,
+      "loss": 0.0043,
+      "step": 19252
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.385689647315984e-05,
+      "loss": 0.0006,
+      "step": 19253
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3844629720124654e-05,
+      "loss": 0.0001,
+      "step": 19254
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.38323649071034e-05,
+      "loss": 0.0054,
+      "step": 19255
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.382010203430094e-05,
+      "loss": 0.0009,
+      "step": 19256
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3807841101922096e-05,
+      "loss": 0.0045,
+      "step": 19257
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3795582110171606e-05,
+      "loss": 0.0059,
+      "step": 19258
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.378332505925427e-05,
+      "loss": 0.0022,
+      "step": 19259
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3771069949374776e-05,
+      "loss": 0.0038,
+      "step": 19260
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.375881678073779e-05,
+      "loss": 0.0013,
+      "step": 19261
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.374656555354801e-05,
+      "loss": 0.0052,
+      "step": 19262
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.373431626801002e-05,
+      "loss": 0.0091,
+      "step": 19263
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.372206892432846e-05,
+      "loss": 0.0036,
+      "step": 19264
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.370982352270775e-05,
+      "loss": 0.0053,
+      "step": 19265
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.369758006335258e-05,
+      "loss": 0.0043,
+      "step": 19266
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3685338546467275e-05,
+      "loss": 0.0045,
+      "step": 19267
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.367309897225648e-05,
+      "loss": 0.0038,
+      "step": 19268
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3660861340924454e-05,
+      "loss": 0.0022,
+      "step": 19269
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.364862565267566e-05,
+      "loss": 0.0015,
+      "step": 19270
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3636391907714426e-05,
+      "loss": 0.001,
+      "step": 19271
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.362416010624513e-05,
+      "loss": 0.0049,
+      "step": 19272
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.361193024847202e-05,
+      "loss": 0.0003,
+      "step": 19273
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.359970233459937e-05,
+      "loss": 0.006,
+      "step": 19274
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3587476364831406e-05,
+      "loss": 0.0059,
+      "step": 19275
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.357525233937234e-05,
+      "loss": 0.0075,
+      "step": 19276
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.356303025842636e-05,
+      "loss": 0.0057,
+      "step": 19277
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.355081012219749e-05,
+      "loss": 0.0034,
+      "step": 19278
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.353859193088999e-05,
+      "loss": 0.0043,
+      "step": 19279
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3526375684707766e-05,
+      "loss": 0.0032,
+      "step": 19280
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3514161383854994e-05,
+      "loss": 0.0126,
+      "step": 19281
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.350194902853557e-05,
+      "loss": 0.0048,
+      "step": 19282
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3489738618953466e-05,
+      "loss": 0.005,
+      "step": 19283
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3477530155312735e-05,
+      "loss": 0.0024,
+      "step": 19284
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.346532363781714e-05,
+      "loss": 0.0023,
+      "step": 19285
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3453119066670667e-05,
+      "loss": 0.0002,
+      "step": 19286
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3440916442077084e-05,
+      "loss": 0.0019,
+      "step": 19287
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.342871576424021e-05,
+      "loss": 0.0049,
+      "step": 19288
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.341651703336383e-05,
+      "loss": 0.0025,
+      "step": 19289
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3404320249651675e-05,
+      "loss": 0.0011,
+      "step": 19290
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.339212541330745e-05,
+      "loss": 0.0086,
+      "step": 19291
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3379932524534866e-05,
+      "loss": 0.0029,
+      "step": 19292
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3367741583537526e-05,
+      "loss": 0.012,
+      "step": 19293
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.335555259051908e-05,
+      "loss": 0.02,
+      "step": 19294
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.334336554568308e-05,
+      "loss": 0.0043,
+      "step": 19295
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.333118044923309e-05,
+      "loss": 0.008,
+      "step": 19296
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.331899730137263e-05,
+      "loss": 0.0015,
+      "step": 19297
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.33068161023051e-05,
+      "loss": 0.0006,
+      "step": 19298
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3294636852234105e-05,
+      "loss": 0.0024,
+      "step": 19299
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.328245955136287e-05,
+      "loss": 0.0053,
+      "step": 19300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.327028419989498e-05,
+      "loss": 0.0022,
+      "step": 19301
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.325811079803363e-05,
+      "loss": 0.0077,
+      "step": 19302
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3245939345982215e-05,
+      "loss": 0.0016,
+      "step": 19303
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.323376984394398e-05,
+      "loss": 0.0068,
+      "step": 19304
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3221602292122215e-05,
+      "loss": 0.005,
+      "step": 19305
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.320943669072013e-05,
+      "loss": 0.0046,
+      "step": 19306
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.319727303994091e-05,
+      "loss": 0.0013,
+      "step": 19307
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.31851113399877e-05,
+      "loss": 0.002,
+      "step": 19308
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.317295159106364e-05,
+      "loss": 0.0088,
+      "step": 19309
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.316079379337184e-05,
+      "loss": 0.0029,
+      "step": 19310
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.314863794711528e-05,
+      "loss": 0.005,
+      "step": 19311
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.313648405249707e-05,
+      "loss": 0.0002,
+      "step": 19312
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.312433210972017e-05,
+      "loss": 0.0002,
+      "step": 19313
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.311218211898754e-05,
+      "loss": 0.0045,
+      "step": 19314
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.310003408050214e-05,
+      "loss": 0.0072,
+      "step": 19315
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3087887994466773e-05,
+      "loss": 0.0018,
+      "step": 19316
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.307574386108443e-05,
+      "loss": 0.0033,
+      "step": 19317
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.30636016805578e-05,
+      "loss": 0.0009,
+      "step": 19318
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3051461453089836e-05,
+      "loss": 0.0013,
+      "step": 19319
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3039323178883184e-05,
+      "loss": 0.024,
+      "step": 19320
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3027186858140615e-05,
+      "loss": 0.004,
+      "step": 19321
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.301505249106482e-05,
+      "loss": 0.0019,
+      "step": 19322
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.300292007785847e-05,
+      "loss": 0.0075,
+      "step": 19323
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.2990789618724214e-05,
+      "loss": 0.0025,
+      "step": 19324
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.2978661113864637e-05,
+      "loss": 0.0011,
+      "step": 19325
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2966534563482315e-05,
+      "loss": 0.0041,
+      "step": 19326
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.295440996777978e-05,
+      "loss": 0.0086,
+      "step": 19327
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.294228732695957e-05,
+      "loss": 0.0018,
+      "step": 19328
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.293016664122405e-05,
+      "loss": 0.0046,
+      "step": 19329
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2918047910775816e-05,
+      "loss": 0.0004,
+      "step": 19330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.29059311358171e-05,
+      "loss": 0.0022,
+      "step": 19331
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.289381631655047e-05,
+      "loss": 0.0003,
+      "step": 19332
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2881703453178106e-05,
+      "loss": 0.0027,
+      "step": 19333
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.286959254590236e-05,
+      "loss": 0.0039,
+      "step": 19334
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2857483594925525e-05,
+      "loss": 0.0021,
+      "step": 19335
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.28453766004498e-05,
+      "loss": 0.0056,
+      "step": 19336
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.283327156267751e-05,
+      "loss": 0.0036,
+      "step": 19337
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.282116848181071e-05,
+      "loss": 0.0007,
+      "step": 19338
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2809067358051574e-05,
+      "loss": 0.0059,
+      "step": 19339
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2796968191602245e-05,
+      "loss": 0.0103,
+      "step": 19340
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2784870982664785e-05,
+      "loss": 0.0046,
+      "step": 19341
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.277277573144123e-05,
+      "loss": 0.0114,
+      "step": 19342
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.276068243813361e-05,
+      "loss": 0.0022,
+      "step": 19343
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.274859110294388e-05,
+      "loss": 0.0006,
+      "step": 19344
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.273650172607403e-05,
+      "loss": 0.001,
+      "step": 19345
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2724414307725944e-05,
+      "loss": 0.0066,
+      "step": 19346
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.27123288481015e-05,
+      "loss": 0.0044,
+      "step": 19347
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.270024534740262e-05,
+      "loss": 0.0053,
+      "step": 19348
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.268816380583096e-05,
+      "loss": 0.0005,
+      "step": 19349
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.267608422358851e-05,
+      "loss": 0.001,
+      "step": 19350
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.266400660087685e-05,
+      "loss": 0.0052,
+      "step": 19351
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.265193093789779e-05,
+      "loss": 0.0002,
+      "step": 19352
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2639857234853e-05,
+      "loss": 0.0041,
+      "step": 19353
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.26277854919441e-05,
+      "loss": 0.0106,
+      "step": 19354
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2615715709372754e-05,
+      "loss": 0.0072,
+      "step": 19355
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.260364788734053e-05,
+      "loss": 0.0034,
+      "step": 19356
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2591582026048986e-05,
+      "loss": 0.0037,
+      "step": 19357
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.257951812569966e-05,
+      "loss": 0.0004,
+      "step": 19358
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.256745618649402e-05,
+      "loss": 0.0024,
+      "step": 19359
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2555396208633525e-05,
+      "loss": 0.0095,
+      "step": 19360
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2543338192319645e-05,
+      "loss": 0.0096,
+      "step": 19361
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.253128213775366e-05,
+      "loss": 0.0075,
+      "step": 19362
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.251922804513708e-05,
+      "loss": 0.005,
+      "step": 19363
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2507175914671074e-05,
+      "loss": 0.0061,
+      "step": 19364
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.249512574655705e-05,
+      "loss": 0.003,
+      "step": 19365
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.248307754099626e-05,
+      "loss": 0.0009,
+      "step": 19366
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.247103129818985e-05,
+      "loss": 0.0008,
+      "step": 19367
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.245898701833913e-05,
+      "loss": 0.0025,
+      "step": 19368
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2446944701645125e-05,
+      "loss": 0.0003,
+      "step": 19369
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2434904348309124e-05,
+      "loss": 0.0056,
+      "step": 19370
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2422865958532104e-05,
+      "loss": 0.0024,
+      "step": 19371
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.241082953251515e-05,
+      "loss": 0.0051,
+      "step": 19372
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2398795070459316e-05,
+      "loss": 0.0032,
+      "step": 19373
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.23867625725656e-05,
+      "loss": 0.002,
+      "step": 19374
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.237473203903496e-05,
+      "loss": 0.0083,
+      "step": 19375
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.236270347006832e-05,
+      "loss": 0.0029,
+      "step": 19376
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2350676865866585e-05,
+      "loss": 0.0028,
+      "step": 19377
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.233865222663064e-05,
+      "loss": 0.0017,
+      "step": 19378
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.232662955256132e-05,
+      "loss": 0.0047,
+      "step": 19379
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.231460884385936e-05,
+      "loss": 0.0003,
+      "step": 19380
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.230259010072563e-05,
+      "loss": 0.0033,
+      "step": 19381
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2290573323360755e-05,
+      "loss": 0.0022,
+      "step": 19382
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.227855851196557e-05,
+      "loss": 0.0045,
+      "step": 19383
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.226654566674064e-05,
+      "loss": 0.0024,
+      "step": 19384
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.225453478788663e-05,
+      "loss": 0.0004,
+      "step": 19385
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.224252587560415e-05,
+      "loss": 0.0037,
+      "step": 19386
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2230518930093776e-05,
+      "loss": 0.0042,
+      "step": 19387
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.221851395155603e-05,
+      "loss": 0.0096,
+      "step": 19388
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2206510940191445e-05,
+      "loss": 0.0044,
+      "step": 19389
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2194509896200455e-05,
+      "loss": 0.0053,
+      "step": 19390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.218251081978354e-05,
+      "loss": 0.0003,
+      "step": 19391
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2170513711141095e-05,
+      "loss": 0.003,
+      "step": 19392
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.215851857047349e-05,
+      "loss": 0.0002,
+      "step": 19393
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.214652539798108e-05,
+      "loss": 0.0026,
+      "step": 19394
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.213453419386411e-05,
+      "loss": 0.0021,
+      "step": 19395
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2122544958322936e-05,
+      "loss": 0.0017,
+      "step": 19396
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.211055769155776e-05,
+      "loss": 0.0035,
+      "step": 19397
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2098572393768813e-05,
+      "loss": 0.0187,
+      "step": 19398
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2086589065156304e-05,
+      "loss": 0.0013,
+      "step": 19399
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.207460770592026e-05,
+      "loss": 0.001,
+      "step": 19400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.206262831626093e-05,
+      "loss": 0.0002,
+      "step": 19401
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.205065089637829e-05,
+      "loss": 0.0015,
+      "step": 19402
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.203867544647244e-05,
+      "loss": 0.0085,
+      "step": 19403
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2026701966743374e-05,
+      "loss": 0.0001,
+      "step": 19404
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.201473045739107e-05,
+      "loss": 0.0092,
+      "step": 19405
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.200276091861548e-05,
+      "loss": 0.0058,
+      "step": 19406
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.199079335061652e-05,
+      "loss": 0.0031,
+      "step": 19407
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.197882775359408e-05,
+      "loss": 0.0002,
+      "step": 19408
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1966864127747987e-05,
+      "loss": 0.0008,
+      "step": 19409
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1954902473278074e-05,
+      "loss": 0.0093,
+      "step": 19410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.19429427903841e-05,
+      "loss": 0.0049,
+      "step": 19411
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.193098507926589e-05,
+      "loss": 0.0001,
+      "step": 19412
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.191902934012301e-05,
+      "loss": 0.0026,
+      "step": 19413
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.190707557315531e-05,
+      "loss": 0.0104,
+      "step": 19414
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.189512377856228e-05,
+      "loss": 0.0011,
+      "step": 19415
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.188317395654371e-05,
+      "loss": 0.0019,
+      "step": 19416
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1871226107299056e-05,
+      "loss": 0.0015,
+      "step": 19417
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.185928023102788e-05,
+      "loss": 0.0011,
+      "step": 19418
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.184733632792979e-05,
+      "loss": 0.0125,
+      "step": 19419
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1835394398204134e-05,
+      "loss": 0.0018,
+      "step": 19420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1823454442050536e-05,
+      "loss": 0.0009,
+      "step": 19421
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1811516459668276e-05,
+      "loss": 0.0067,
+      "step": 19422
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1799580451256794e-05,
+      "loss": 0.0037,
+      "step": 19423
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.178764641701544e-05,
+      "loss": 0.0023,
+      "step": 19424
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1775714357143514e-05,
+      "loss": 0.0079,
+      "step": 19425
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1763784271840336e-05,
+      "loss": 0.002,
+      "step": 19426
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.175185616130514e-05,
+      "loss": 0.0013,
+      "step": 19427
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1739930025737163e-05,
+      "loss": 0.0024,
+      "step": 19428
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1728005865335587e-05,
+      "loss": 0.0094,
+      "step": 19429
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.17160836802996e-05,
+      "loss": 0.0006,
+      "step": 19430
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.170416347082822e-05,
+      "loss": 0.002,
+      "step": 19431
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.16922452371207e-05,
+      "loss": 0.0061,
+      "step": 19432
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1680328979375895e-05,
+      "loss": 0.0098,
+      "step": 19433
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.166841469779305e-05,
+      "loss": 0.0023,
+      "step": 19434
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1656502392570975e-05,
+      "loss": 0.0011,
+      "step": 19435
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.164459206390872e-05,
+      "loss": 0.0039,
+      "step": 19436
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.163268371200519e-05,
+      "loss": 0.0018,
+      "step": 19437
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.162077733705928e-05,
+      "loss": 0.0054,
+      "step": 19438
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.160887293926983e-05,
+      "loss": 0.0003,
+      "step": 19439
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.159697051883569e-05,
+      "loss": 0.0016,
+      "step": 19440
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1585070075955635e-05,
+      "loss": 0.0058,
+      "step": 19441
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.157317161082845e-05,
+      "loss": 0.0003,
+      "step": 19442
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1561275123652835e-05,
+      "loss": 0.0162,
+      "step": 19443
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1549380614627495e-05,
+      "loss": 0.0014,
+      "step": 19444
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.153748808395114e-05,
+      "loss": 0.0012,
+      "step": 19445
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.152559753182226e-05,
+      "loss": 0.0022,
+      "step": 19446
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.151370895843962e-05,
+      "loss": 0.003,
+      "step": 19447
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.150182236400161e-05,
+      "loss": 0.0012,
+      "step": 19448
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.148993774870689e-05,
+      "loss": 0.0105,
+      "step": 19449
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.147805511275396e-05,
+      "loss": 0.0038,
+      "step": 19450
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1466174456341144e-05,
+      "loss": 0.0003,
+      "step": 19451
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.145429577966706e-05,
+      "loss": 0.0038,
+      "step": 19452
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1442419082929954e-05,
+      "loss": 0.0011,
+      "step": 19453
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.143054436632825e-05,
+      "loss": 0.0095,
+      "step": 19454
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.141867163006027e-05,
+      "loss": 0.0027,
+      "step": 19455
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1406800874324314e-05,
+      "loss": 0.001,
+      "step": 19456
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1394932099318646e-05,
+      "loss": 0.0058,
+      "step": 19457
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1383065305241474e-05,
+      "loss": 0.0033,
+      "step": 19458
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.137120049229103e-05,
+      "loss": 0.0065,
+      "step": 19459
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1359337660665476e-05,
+      "loss": 0.0013,
+      "step": 19460
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.134747681056292e-05,
+      "loss": 0.0058,
+      "step": 19461
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1335617942181485e-05,
+      "loss": 0.0075,
+      "step": 19462
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.132376105571926e-05,
+      "loss": 0.0045,
+      "step": 19463
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1311906151374156e-05,
+      "loss": 0.028,
+      "step": 19464
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.130005322934435e-05,
+      "loss": 0.0022,
+      "step": 19465
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1288202289827626e-05,
+      "loss": 0.0065,
+      "step": 19466
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.12763533330221e-05,
+      "loss": 0.0035,
+      "step": 19467
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1264506359125525e-05,
+      "loss": 0.0031,
+      "step": 19468
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.125266136833583e-05,
+      "loss": 0.0015,
+      "step": 19469
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.124081836085084e-05,
+      "loss": 0.0074,
+      "step": 19470
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.122897733686833e-05,
+      "loss": 0.0039,
+      "step": 19471
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1217138296586154e-05,
+      "loss": 0.0036,
+      "step": 19472
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1205301240201956e-05,
+      "loss": 0.0018,
+      "step": 19473
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1193466167913466e-05,
+      "loss": 0.0027,
+      "step": 19474
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1181633079918375e-05,
+      "loss": 0.0043,
+      "step": 19475
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.116980197641431e-05,
+      "loss": 0.0031,
+      "step": 19476
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.115797285759879e-05,
+      "loss": 0.0001,
+      "step": 19477
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.11461457236695e-05,
+      "loss": 0.0008,
+      "step": 19478
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.113432057482391e-05,
+      "loss": 0.0016,
+      "step": 19479
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.112249741125956e-05,
+      "loss": 0.0019,
+      "step": 19480
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.111067623317394e-05,
+      "loss": 0.0024,
+      "step": 19481
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1098857040764366e-05,
+      "loss": 0.0021,
+      "step": 19482
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1087039834228386e-05,
+      "loss": 0.0108,
+      "step": 19483
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.107522461376323e-05,
+      "loss": 0.0014,
+      "step": 19484
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.10634113795664e-05,
+      "loss": 0.0046,
+      "step": 19485
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.105160013183506e-05,
+      "loss": 0.0421,
+      "step": 19486
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1039790870766526e-05,
+      "loss": 0.0009,
+      "step": 19487
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1027983596558056e-05,
+      "loss": 0.0033,
+      "step": 19488
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.101617830940683e-05,
+      "loss": 0.0063,
+      "step": 19489
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.100437500951e-05,
+      "loss": 0.0005,
+      "step": 19490
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0992573697064746e-05,
+      "loss": 0.0028,
+      "step": 19491
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.098077437226816e-05,
+      "loss": 0.0028,
+      "step": 19492
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.096897703531729e-05,
+      "loss": 0.0024,
+      "step": 19493
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.09571816864092e-05,
+      "loss": 0.0088,
+      "step": 19494
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.094538832574088e-05,
+      "loss": 0.0014,
+      "step": 19495
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.093359695350937e-05,
+      "loss": 0.0005,
+      "step": 19496
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.092180756991144e-05,
+      "loss": 0.0081,
+      "step": 19497
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0910020175144176e-05,
+      "loss": 0.0029,
+      "step": 19498
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0898234769404344e-05,
+      "loss": 0.0022,
+      "step": 19499
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0886451352888816e-05,
+      "loss": 0.0049,
+      "step": 19500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0874669925794393e-05,
+      "loss": 0.0049,
+      "step": 19501
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.086289048831779e-05,
+      "loss": 0.0046,
+      "step": 19502
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.08511130406559e-05,
+      "loss": 0.0045,
+      "step": 19503
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.083933758300529e-05,
+      "loss": 0.0163,
+      "step": 19504
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.082756411556266e-05,
+      "loss": 0.0117,
+      "step": 19505
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.081579263852468e-05,
+      "loss": 0.0006,
+      "step": 19506
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0804023152087925e-05,
+      "loss": 0.0001,
+      "step": 19507
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.079225565644899e-05,
+      "loss": 0.0046,
+      "step": 19508
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.078049015180442e-05,
+      "loss": 0.0068,
+      "step": 19509
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.076872663835068e-05,
+      "loss": 0.0004,
+      "step": 19510
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.075696511628429e-05,
+      "loss": 0.001,
+      "step": 19511
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0745205585801656e-05,
+      "loss": 0.0017,
+      "step": 19512
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.073344804709922e-05,
+      "loss": 0.0004,
+      "step": 19513
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.072169250037335e-05,
+      "loss": 0.0004,
+      "step": 19514
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.07099389458203e-05,
+      "loss": 0.0027,
+      "step": 19515
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0698187383636524e-05,
+      "loss": 0.0067,
+      "step": 19516
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.068643781401813e-05,
+      "loss": 0.0016,
+      "step": 19517
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.067469023716154e-05,
+      "loss": 0.0089,
+      "step": 19518
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.066294465326281e-05,
+      "loss": 0.0054,
+      "step": 19519
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.065120106251818e-05,
+      "loss": 0.0009,
+      "step": 19520
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.063945946512376e-05,
+      "loss": 0.0006,
+      "step": 19521
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.062771986127569e-05,
+      "loss": 0.0018,
+      "step": 19522
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.061598225117004e-05,
+      "loss": 0.0011,
+      "step": 19523
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.060424663500282e-05,
+      "loss": 0.0035,
+      "step": 19524
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.059251301297006e-05,
+      "loss": 0.0052,
+      "step": 19525
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.058078138526775e-05,
+      "loss": 0.044,
+      "step": 19526
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0569051752091834e-05,
+      "loss": 0.0045,
+      "step": 19527
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.055732411363812e-05,
+      "loss": 0.0003,
+      "step": 19528
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0545598470102624e-05,
+      "loss": 0.0049,
+      "step": 19529
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0533874821681046e-05,
+      "loss": 0.0031,
+      "step": 19530
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.052215316856932e-05,
+      "loss": 0.0017,
+      "step": 19531
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0510433510963195e-05,
+      "loss": 0.0147,
+      "step": 19532
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0498715849058302e-05,
+      "loss": 0.009,
+      "step": 19533
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0487000183050526e-05,
+      "loss": 0.0034,
+      "step": 19534
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.047528651313534e-05,
+      "loss": 0.0079,
+      "step": 19535
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0463574839508587e-05,
+      "loss": 0.0058,
+      "step": 19536
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.045186516236573e-05,
+      "loss": 0.0064,
+      "step": 19537
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0440157481902376e-05,
+      "loss": 0.0047,
+      "step": 19538
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.042845179831409e-05,
+      "loss": 0.0134,
+      "step": 19539
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0416748111796363e-05,
+      "loss": 0.0105,
+      "step": 19540
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.040504642254467e-05,
+      "loss": 0.0075,
+      "step": 19541
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0393346730754458e-05,
+      "loss": 0.0117,
+      "step": 19542
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.038164903662112e-05,
+      "loss": 0.0017,
+      "step": 19543
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0369953340340048e-05,
+      "loss": 0.0041,
+      "step": 19544
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0358259642106568e-05,
+      "loss": 0.008,
+      "step": 19545
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0346567942115995e-05,
+      "loss": 0.0023,
+      "step": 19546
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.033487824056364e-05,
+      "loss": 0.0019,
+      "step": 19547
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.032319053764462e-05,
+      "loss": 0.0032,
+      "step": 19548
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0311504833554305e-05,
+      "loss": 0.0009,
+      "step": 19549
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.029982112848774e-05,
+      "loss": 0.0051,
+      "step": 19550
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.028813942264013e-05,
+      "loss": 0.0024,
+      "step": 19551
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0276459716206535e-05,
+      "loss": 0.0085,
+      "step": 19552
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.026478200938209e-05,
+      "loss": 0.0058,
+      "step": 19553
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.025310630236178e-05,
+      "loss": 0.0036,
+      "step": 19554
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0241432595340636e-05,
+      "loss": 0.0032,
+      "step": 19555
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.022976088851364e-05,
+      "loss": 0.0003,
+      "step": 19556
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.021809118207569e-05,
+      "loss": 0.0015,
+      "step": 19557
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0206423476221746e-05,
+      "loss": 0.0036,
+      "step": 19558
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.019475777114666e-05,
+      "loss": 0.0179,
+      "step": 19559
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0183094067045293e-05,
+      "loss": 0.0038,
+      "step": 19560
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0171432364112357e-05,
+      "loss": 0.0033,
+      "step": 19561
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.015977266254273e-05,
+      "loss": 0.0022,
+      "step": 19562
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0148114962531106e-05,
+      "loss": 0.0032,
+      "step": 19563
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0136459264272213e-05,
+      "loss": 0.0038,
+      "step": 19564
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.012480556796074e-05,
+      "loss": 0.0013,
+      "step": 19565
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0113153873791206e-05,
+      "loss": 0.0041,
+      "step": 19566
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0101504181958402e-05,
+      "loss": 0.0019,
+      "step": 19567
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.008985649265672e-05,
+      "loss": 0.0062,
+      "step": 19568
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0078210806080842e-05,
+      "loss": 0.0058,
+      "step": 19569
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0066567122425196e-05,
+      "loss": 0.0028,
+      "step": 19570
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0054925441884252e-05,
+      "loss": 0.0035,
+      "step": 19571
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.004328576465248e-05,
+      "loss": 0.0051,
+      "step": 19572
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0031648090924254e-05,
+      "loss": 0.012,
+      "step": 19573
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0020012420893968e-05,
+      "loss": 0.0042,
+      "step": 19574
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0008378754755967e-05,
+      "loss": 0.0007,
+      "step": 19575
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9996747092704516e-05,
+      "loss": 0.003,
+      "step": 19576
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9985117434933937e-05,
+      "loss": 0.0036,
+      "step": 19577
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.997348978163846e-05,
+      "loss": 0.0043,
+      "step": 19578
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.99618641330122e-05,
+      "loss": 0.0017,
+      "step": 19579
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.995024048924948e-05,
+      "loss": 0.0031,
+      "step": 19580
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.993861885054426e-05,
+      "loss": 0.0065,
+      "step": 19581
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9926999217090825e-05,
+      "loss": 0.0047,
+      "step": 19582
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9915381589083113e-05,
+      "loss": 0.0047,
+      "step": 19583
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9903765966715176e-05,
+      "loss": 0.0014,
+      "step": 19584
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.989215235018112e-05,
+      "loss": 0.0015,
+      "step": 19585
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9880540739674742e-05,
+      "loss": 0.0011,
+      "step": 19586
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9868931135390183e-05,
+      "loss": 0.0103,
+      "step": 19587
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9857323537521178e-05,
+      "loss": 0.0017,
+      "step": 19588
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9845717946261665e-05,
+      "loss": 0.0209,
+      "step": 19589
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9834114361805463e-05,
+      "loss": 0.0023,
+      "step": 19590
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9822512784346374e-05,
+      "loss": 0.002,
+      "step": 19591
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9810913214078187e-05,
+      "loss": 0.0099,
+      "step": 19592
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9799315651194622e-05,
+      "loss": 0.0034,
+      "step": 19593
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.978772009588938e-05,
+      "loss": 0.004,
+      "step": 19594
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9776126548356115e-05,
+      "loss": 0.0015,
+      "step": 19595
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9764535008788497e-05,
+      "loss": 0.0004,
+      "step": 19596
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9752945477380086e-05,
+      "loss": 0.0051,
+      "step": 19597
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.974135795432451e-05,
+      "loss": 0.0012,
+      "step": 19598
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.972977243981519e-05,
+      "loss": 0.0031,
+      "step": 19599
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9718188934045765e-05,
+      "loss": 0.0011,
+      "step": 19600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9706607437209612e-05,
+      "loss": 0.0081,
+      "step": 19601
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9695027949500173e-05,
+      "loss": 0.001,
+      "step": 19602
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.968345047111086e-05,
+      "loss": 0.0022,
+      "step": 19603
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9671875002235033e-05,
+      "loss": 0.0007,
+      "step": 19604
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9660301543066027e-05,
+      "loss": 0.0076,
+      "step": 19605
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.964873009379715e-05,
+      "loss": 0.0033,
+      "step": 19606
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.963716065462165e-05,
+      "loss": 0.0019,
+      "step": 19607
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9625593225732787e-05,
+      "loss": 0.0017,
+      "step": 19608
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.961402780732372e-05,
+      "loss": 0.0068,
+      "step": 19609
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.960246439958765e-05,
+      "loss": 0.013,
+      "step": 19610
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.959090300271772e-05,
+      "loss": 0.0063,
+      "step": 19611
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.957934361690691e-05,
+      "loss": 0.0029,
+      "step": 19612
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9567786242348462e-05,
+      "loss": 0.0016,
+      "step": 19613
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9556230879235245e-05,
+      "loss": 0.0033,
+      "step": 19614
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9544677527760346e-05,
+      "loss": 0.0043,
+      "step": 19615
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.953312618811677e-05,
+      "loss": 0.0045,
+      "step": 19616
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9521576860497288e-05,
+      "loss": 0.0035,
+      "step": 19617
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.951002954509498e-05,
+      "loss": 0.0003,
+      "step": 19618
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9498484242102533e-05,
+      "loss": 0.0129,
+      "step": 19619
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.948694095171295e-05,
+      "loss": 0.0067,
+      "step": 19620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9475399674118882e-05,
+      "loss": 0.0123,
+      "step": 19621
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9463860409513157e-05,
+      "loss": 0.0037,
+      "step": 19622
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9452323158088487e-05,
+      "loss": 0.0005,
+      "step": 19623
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9440787920037584e-05,
+      "loss": 0.0091,
+      "step": 19624
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.942925469555308e-05,
+      "loss": 0.0016,
+      "step": 19625
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9417723484827616e-05,
+      "loss": 0.0024,
+      "step": 19626
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.940619428805379e-05,
+      "loss": 0.0008,
+      "step": 19627
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.939466710542415e-05,
+      "loss": 0.0043,
+      "step": 19628
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9383141937131278e-05,
+      "loss": 0.0079,
+      "step": 19629
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.937161878336753e-05,
+      "loss": 0.0046,
+      "step": 19630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9360097644325537e-05,
+      "loss": 0.0153,
+      "step": 19631
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9348578520197548e-05,
+      "loss": 0.0072,
+      "step": 19632
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9337061411176122e-05,
+      "loss": 0.0039,
+      "step": 19633
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9325546317453507e-05,
+      "loss": 0.0031,
+      "step": 19634
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.931403323922205e-05,
+      "loss": 0.008,
+      "step": 19635
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9302522176674047e-05,
+      "loss": 0.002,
+      "step": 19636
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.929101313000173e-05,
+      "loss": 0.0025,
+      "step": 19637
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9279506099397422e-05,
+      "loss": 0.0004,
+      "step": 19638
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9268001085053188e-05,
+      "loss": 0.0002,
+      "step": 19639
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9256498087161247e-05,
+      "loss": 0.0027,
+      "step": 19640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9244997105913705e-05,
+      "loss": 0.0014,
+      "step": 19641
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9233498141502643e-05,
+      "loss": 0.0002,
+      "step": 19642
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9222001194120143e-05,
+      "loss": 0.0026,
+      "step": 19643
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9210506263958184e-05,
+      "loss": 0.0017,
+      "step": 19644
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9199013351208793e-05,
+      "loss": 0.0056,
+      "step": 19645
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9187522456063906e-05,
+      "loss": 0.0016,
+      "step": 19646
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.917603357871548e-05,
+      "loss": 0.0006,
+      "step": 19647
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.916454671935528e-05,
+      "loss": 0.0002,
+      "step": 19648
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9153061878175337e-05,
+      "loss": 0.0025,
+      "step": 19649
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9141579055367283e-05,
+      "loss": 0.0015,
+      "step": 19650
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9130098251123075e-05,
+      "loss": 0.0005,
+      "step": 19651
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9118619465634348e-05,
+      "loss": 0.0015,
+      "step": 19652
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9107142699092862e-05,
+      "loss": 0.0011,
+      "step": 19653
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.909566795169028e-05,
+      "loss": 0.0016,
+      "step": 19654
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.908419522361829e-05,
+      "loss": 0.0033,
+      "step": 19655
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9072724515068463e-05,
+      "loss": 0.0028,
+      "step": 19656
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9061255826232416e-05,
+      "loss": 0.0011,
+      "step": 19657
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.904978915730168e-05,
+      "loss": 0.0003,
+      "step": 19658
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9038324508467764e-05,
+      "loss": 0.0005,
+      "step": 19659
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.902686187992217e-05,
+      "loss": 0.0006,
+      "step": 19660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9015401271856325e-05,
+      "loss": 0.0005,
+      "step": 19661
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.900394268446171e-05,
+      "loss": 0.003,
+      "step": 19662
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8992486117929557e-05,
+      "loss": 0.0014,
+      "step": 19663
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8981031572451395e-05,
+      "loss": 0.0021,
+      "step": 19664
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8969579048218354e-05,
+      "loss": 0.0006,
+      "step": 19665
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8958128545421882e-05,
+      "loss": 0.0011,
+      "step": 19666
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8946680064253125e-05,
+      "loss": 0.0018,
+      "step": 19667
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8935233604903264e-05,
+      "loss": 0.0022,
+      "step": 19668
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8923789167563594e-05,
+      "loss": 0.0065,
+      "step": 19669
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.891234675242513e-05,
+      "loss": 0.0058,
+      "step": 19670
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.890090635967912e-05,
+      "loss": 0.0001,
+      "step": 19671
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8889467989516528e-05,
+      "loss": 0.002,
+      "step": 19672
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.887803164212842e-05,
+      "loss": 0.0008,
+      "step": 19673
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8866597317705837e-05,
+      "loss": 0.0016,
+      "step": 19674
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8855165016439712e-05,
+      "loss": 0.0029,
+      "step": 19675
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.884373473852103e-05,
+      "loss": 0.0024,
+      "step": 19676
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8832306484140666e-05,
+      "loss": 0.009,
+      "step": 19677
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8820880253489503e-05,
+      "loss": 0.0007,
+      "step": 19678
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8809456046758372e-05,
+      "loss": 0.0003,
+      "step": 19679
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.879803386413814e-05,
+      "loss": 0.0026,
+      "step": 19680
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8786613705819445e-05,
+      "loss": 0.0012,
+      "step": 19681
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8775195571993192e-05,
+      "loss": 0.0009,
+      "step": 19682
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8763779462849912e-05,
+      "loss": 0.0019,
+      "step": 19683
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.875236537858045e-05,
+      "loss": 0.0028,
+      "step": 19684
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.874095331937532e-05,
+      "loss": 0.0039,
+      "step": 19685
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.872954328542517e-05,
+      "loss": 0.0026,
+      "step": 19686
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.871813527692055e-05,
+      "loss": 0.0005,
+      "step": 19687
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8706729294052004e-05,
+      "loss": 0.0027,
+      "step": 19688
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8695325337010033e-05,
+      "loss": 0.0011,
+      "step": 19689
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8683923405985115e-05,
+      "loss": 0.0011,
+      "step": 19690
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8672523501167684e-05,
+      "loss": 0.0015,
+      "step": 19691
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8661125622748117e-05,
+      "loss": 0.0024,
+      "step": 19692
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8649729770916812e-05,
+      "loss": 0.0024,
+      "step": 19693
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8638335945864073e-05,
+      "loss": 0.0004,
+      "step": 19694
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8626944147780255e-05,
+      "loss": 0.0012,
+      "step": 19695
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8615554376855497e-05,
+      "loss": 0.0001,
+      "step": 19696
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8604166633280144e-05,
+      "loss": 0.0038,
+      "step": 19697
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8592780917244396e-05,
+      "loss": 0.0041,
+      "step": 19698
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8581397228938314e-05,
+      "loss": 0.0003,
+      "step": 19699
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8570015568552164e-05,
+      "loss": 0.0017,
+      "step": 19700
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8558635936275892e-05,
+      "loss": 0.0013,
+      "step": 19701
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8547258332299718e-05,
+      "loss": 0.0031,
+      "step": 19702
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8535882756813533e-05,
+      "loss": 0.0004,
+      "step": 19703
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.852450921000741e-05,
+      "loss": 0.0063,
+      "step": 19704
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.851313769207127e-05,
+      "loss": 0.0005,
+      "step": 19705
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8501768203195053e-05,
+      "loss": 0.0014,
+      "step": 19706
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8490400743568653e-05,
+      "loss": 0.0034,
+      "step": 19707
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.847903531338192e-05,
+      "loss": 0.0036,
+      "step": 19708
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8467671912824703e-05,
+      "loss": 0.0047,
+      "step": 19709
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8456310542086763e-05,
+      "loss": 0.0008,
+      "step": 19710
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.844495120135785e-05,
+      "loss": 0.0018,
+      "step": 19711
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.843359389082773e-05,
+      "loss": 0.0002,
+      "step": 19712
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8422238610686087e-05,
+      "loss": 0.0009,
+      "step": 19713
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8410885361122483e-05,
+      "loss": 0.0013,
+      "step": 19714
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.839953414232668e-05,
+      "loss": 0.0006,
+      "step": 19715
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.838818495448813e-05,
+      "loss": 0.0013,
+      "step": 19716
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8376837797796515e-05,
+      "loss": 0.0024,
+      "step": 19717
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8365492672441265e-05,
+      "loss": 0.0002,
+      "step": 19718
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8354149578611863e-05,
+      "loss": 0.0014,
+      "step": 19719
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8342808516497805e-05,
+      "loss": 0.0023,
+      "step": 19720
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8331469486288493e-05,
+      "loss": 0.0004,
+      "step": 19721
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.832013248817329e-05,
+      "loss": 0.0002,
+      "step": 19722
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8308797522341564e-05,
+      "loss": 0.0029,
+      "step": 19723
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8297464588982645e-05,
+      "loss": 0.0045,
+      "step": 19724
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8286133688285777e-05,
+      "loss": 0.007,
+      "step": 19725
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8274804820440284e-05,
+      "loss": 0.0002,
+      "step": 19726
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8263477985635224e-05,
+      "loss": 0.0022,
+      "step": 19727
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8252153184059933e-05,
+      "loss": 0.003,
+      "step": 19728
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8240830415903492e-05,
+      "loss": 0.001,
+      "step": 19729
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8229509681355016e-05,
+      "loss": 0.0028,
+      "step": 19730
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.821819098060364e-05,
+      "loss": 0.0035,
+      "step": 19731
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8206874313838258e-05,
+      "loss": 0.0016,
+      "step": 19732
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.819555968124807e-05,
+      "loss": 0.0021,
+      "step": 19733
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.818424708302186e-05,
+      "loss": 0.0044,
+      "step": 19734
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.817293651934876e-05,
+      "loss": 0.0006,
+      "step": 19735
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.816162799041755e-05,
+      "loss": 0.0005,
+      "step": 19736
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8150321496417135e-05,
+      "loss": 0.0016,
+      "step": 19737
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8139017037536356e-05,
+      "loss": 0.0015,
+      "step": 19738
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8127714613964015e-05,
+      "loss": 0.0015,
+      "step": 19739
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8116414225888893e-05,
+      "loss": 0.0016,
+      "step": 19740
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.810511587349973e-05,
+      "loss": 0.0004,
+      "step": 19741
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8093819556985214e-05,
+      "loss": 0.0023,
+      "step": 19742
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8082525276534034e-05,
+      "loss": 0.0032,
+      "step": 19743
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8071233032334822e-05,
+      "loss": 0.0022,
+      "step": 19744
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8059942824576176e-05,
+      "loss": 0.0001,
+      "step": 19745
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8048654653446697e-05,
+      "loss": 0.0019,
+      "step": 19746
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8037368519134794e-05,
+      "loss": 0.0018,
+      "step": 19747
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.802608442182916e-05,
+      "loss": 0.0003,
+      "step": 19748
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8014802361718102e-05,
+      "loss": 0.0043,
+      "step": 19749
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8003522338990074e-05,
+      "loss": 0.0108,
+      "step": 19750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7992244353833577e-05,
+      "loss": 0.0018,
+      "step": 19751
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7980968406436838e-05,
+      "loss": 0.0004,
+      "step": 19752
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7969694496988326e-05,
+      "loss": 0.0001,
+      "step": 19753
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.795842262567624e-05,
+      "loss": 0.0035,
+      "step": 19754
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.794715279268885e-05,
+      "loss": 0.0051,
+      "step": 19755
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7935884998214414e-05,
+      "loss": 0.0028,
+      "step": 19756
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7924619242441086e-05,
+      "loss": 0.0095,
+      "step": 19757
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.791335552555707e-05,
+      "loss": 0.0001,
+      "step": 19758
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.790209384775047e-05,
+      "loss": 0.0048,
+      "step": 19759
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7890834209209383e-05,
+      "loss": 0.0015,
+      "step": 19760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.787957661012185e-05,
+      "loss": 0.0037,
+      "step": 19761
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7868321050675907e-05,
+      "loss": 0.0019,
+      "step": 19762
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7857067531059552e-05,
+      "loss": 0.0001,
+      "step": 19763
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.784581605146077e-05,
+      "loss": 0.0007,
+      "step": 19764
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7834566612067356e-05,
+      "loss": 0.0065,
+      "step": 19765
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7823319213067364e-05,
+      "loss": 0.0013,
+      "step": 19766
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7812073854648475e-05,
+      "loss": 0.0033,
+      "step": 19767
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.780083053699869e-05,
+      "loss": 0.0006,
+      "step": 19768
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.778958926030566e-05,
+      "loss": 0.0017,
+      "step": 19769
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7778350024757185e-05,
+      "loss": 0.0014,
+      "step": 19770
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.776711283054098e-05,
+      "loss": 0.0003,
+      "step": 19771
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7755877677844706e-05,
+      "loss": 0.0013,
+      "step": 19772
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7744644566856023e-05,
+      "loss": 0.0033,
+      "step": 19773
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7733413497762557e-05,
+      "loss": 0.0007,
+      "step": 19774
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7722184470751896e-05,
+      "loss": 0.0033,
+      "step": 19775
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7710957486011552e-05,
+      "loss": 0.0017,
+      "step": 19776
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7699732543729096e-05,
+      "loss": 0.0021,
+      "step": 19777
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7688509644091895e-05,
+      "loss": 0.0027,
+      "step": 19778
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7677288787287526e-05,
+      "loss": 0.0003,
+      "step": 19779
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.766606997350326e-05,
+      "loss": 0.0004,
+      "step": 19780
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7654853202926587e-05,
+      "loss": 0.0006,
+      "step": 19781
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7643638475744862e-05,
+      "loss": 0.001,
+      "step": 19782
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.763242579214524e-05,
+      "loss": 0.0021,
+      "step": 19783
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.762121515231519e-05,
+      "loss": 0.0003,
+      "step": 19784
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7610006556441762e-05,
+      "loss": 0.0002,
+      "step": 19785
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7598800004712336e-05,
+      "loss": 0.0047,
+      "step": 19786
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7587595497313948e-05,
+      "loss": 0.0003,
+      "step": 19787
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7576393034433797e-05,
+      "loss": 0.0005,
+      "step": 19788
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.756519261625897e-05,
+      "loss": 0.0001,
+      "step": 19789
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7553994242976528e-05,
+      "loss": 0.0018,
+      "step": 19790
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7542797914773528e-05,
+      "loss": 0.0018,
+      "step": 19791
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.753160363183695e-05,
+      "loss": 0.0004,
+      "step": 19792
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.752041139435378e-05,
+      "loss": 0.0001,
+      "step": 19793
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7509221202510915e-05,
+      "loss": 0.0001,
+      "step": 19794
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.749803305649533e-05,
+      "loss": 0.0004,
+      "step": 19795
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7486846956493746e-05,
+      "loss": 0.0037,
+      "step": 19796
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7475662902693174e-05,
+      "loss": 0.0023,
+      "step": 19797
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.746448089528021e-05,
+      "loss": 0.0006,
+      "step": 19798
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7453300934441808e-05,
+      "loss": 0.001,
+      "step": 19799
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7442123020364565e-05,
+      "loss": 0.0025,
+      "step": 19800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.74309471532352e-05,
+      "loss": 0.0005,
+      "step": 19801
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7419773333240397e-05,
+      "loss": 0.0029,
+      "step": 19802
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7408601560566757e-05,
+      "loss": 0.003,
+      "step": 19803
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.739743183540088e-05,
+      "loss": 0.0003,
+      "step": 19804
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7386264157929317e-05,
+      "loss": 0.0056,
+      "step": 19805
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7375098528338603e-05,
+      "loss": 0.0058,
+      "step": 19806
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.73639349468152e-05,
+      "loss": 0.0028,
+      "step": 19807
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7352773413545588e-05,
+      "loss": 0.0003,
+      "step": 19808
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.734161392871617e-05,
+      "loss": 0.0016,
+      "step": 19809
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7330456492513374e-05,
+      "loss": 0.0027,
+      "step": 19810
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.731930110512343e-05,
+      "loss": 0.0013,
+      "step": 19811
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7308147766732773e-05,
+      "loss": 0.0001,
+      "step": 19812
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.729699647752764e-05,
+      "loss": 0.0054,
+      "step": 19813
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.728584723769429e-05,
+      "loss": 0.0017,
+      "step": 19814
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.727470004741897e-05,
+      "loss": 0.0001,
+      "step": 19815
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7263554906887736e-05,
+      "loss": 0.0028,
+      "step": 19816
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7252411816286895e-05,
+      "loss": 0.0008,
+      "step": 19817
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7241270775802414e-05,
+      "loss": 0.0038,
+      "step": 19818
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7230131785620497e-05,
+      "loss": 0.0046,
+      "step": 19819
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.721899484592709e-05,
+      "loss": 0.0014,
+      "step": 19820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7207859956908247e-05,
+      "loss": 0.0033,
+      "step": 19821
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7196727118749917e-05,
+      "loss": 0.002,
+      "step": 19822
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7185596331638053e-05,
+      "loss": 0.0005,
+      "step": 19823
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.717446759575856e-05,
+      "loss": 0.0002,
+      "step": 19824
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.71633409112973e-05,
+      "loss": 0.0034,
+      "step": 19825
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7152216278440124e-05,
+      "loss": 0.0084,
+      "step": 19826
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7141093697372824e-05,
+      "loss": 0.0007,
+      "step": 19827
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.712997316828121e-05,
+      "loss": 0.0046,
+      "step": 19828
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7118854691350907e-05,
+      "loss": 0.0054,
+      "step": 19829
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.710773826676776e-05,
+      "loss": 0.0061,
+      "step": 19830
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.709662389471729e-05,
+      "loss": 0.0085,
+      "step": 19831
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7085511575385266e-05,
+      "loss": 0.006,
+      "step": 19832
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.707440130895717e-05,
+      "loss": 0.0008,
+      "step": 19833
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7063293095618578e-05,
+      "loss": 0.0007,
+      "step": 19834
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.705218693555513e-05,
+      "loss": 0.0023,
+      "step": 19835
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.704108282895217e-05,
+      "loss": 0.0002,
+      "step": 19836
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7029980775995298e-05,
+      "loss": 0.0026,
+      "step": 19837
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.701888077686983e-05,
+      "loss": 0.0017,
+      "step": 19838
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7007782831761197e-05,
+      "loss": 0.0023,
+      "step": 19839
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6996686940854762e-05,
+      "loss": 0.0002,
+      "step": 19840
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6985593104335833e-05,
+      "loss": 0.0014,
+      "step": 19841
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.69745013223897e-05,
+      "loss": 0.0016,
+      "step": 19842
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6963411595201633e-05,
+      "loss": 0.0025,
+      "step": 19843
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6952323922956846e-05,
+      "loss": 0.0026,
+      "step": 19844
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6941238305840496e-05,
+      "loss": 0.003,
+      "step": 19845
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6930154744037813e-05,
+      "loss": 0.0089,
+      "step": 19846
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6919073237733767e-05,
+      "loss": 0.0006,
+      "step": 19847
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6907993787113614e-05,
+      "loss": 0.0001,
+      "step": 19848
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.689691639236225e-05,
+      "loss": 0.0035,
+      "step": 19849
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6885841053664814e-05,
+      "loss": 0.0049,
+      "step": 19850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6874767771206207e-05,
+      "loss": 0.0003,
+      "step": 19851
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.686369654517138e-05,
+      "loss": 0.0015,
+      "step": 19852
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6852627375745257e-05,
+      "loss": 0.0008,
+      "step": 19853
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6841560263112717e-05,
+      "loss": 0.0002,
+      "step": 19854
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6830495207458596e-05,
+      "loss": 0.0016,
+      "step": 19855
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6819432208967718e-05,
+      "loss": 0.0014,
+      "step": 19856
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.680837126782483e-05,
+      "loss": 0.0012,
+      "step": 19857
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6797312384214685e-05,
+      "loss": 0.0004,
+      "step": 19858
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6786255558321972e-05,
+      "loss": 0.0049,
+      "step": 19859
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.677520079033139e-05,
+      "loss": 0.0029,
+      "step": 19860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6764148080427596e-05,
+      "loss": 0.0016,
+      "step": 19861
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6753097428795066e-05,
+      "loss": 0.0025,
+      "step": 19862
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6742048835618545e-05,
+      "loss": 0.0015,
+      "step": 19863
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6731002301082382e-05,
+      "loss": 0.0003,
+      "step": 19864
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6719957825371212e-05,
+      "loss": 0.0036,
+      "step": 19865
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.670891540866949e-05,
+      "loss": 0.0011,
+      "step": 19866
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6697875051161548e-05,
+      "loss": 0.0013,
+      "step": 19867
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6686836753031892e-05,
+      "loss": 0.0007,
+      "step": 19868
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6675800514464806e-05,
+      "loss": 0.0065,
+      "step": 19869
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6664766335644662e-05,
+      "loss": 0.0058,
+      "step": 19870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6653734216755706e-05,
+      "loss": 0.002,
+      "step": 19871
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6642704157982232e-05,
+      "loss": 0.0006,
+      "step": 19872
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6631676159508453e-05,
+      "loss": 0.0018,
+      "step": 19873
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.662065022151854e-05,
+      "loss": 0.0034,
+      "step": 19874
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6609626344196684e-05,
+      "loss": 0.0001,
+      "step": 19875
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.659860452772698e-05,
+      "loss": 0.0017,
+      "step": 19876
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6587584772293503e-05,
+      "loss": 0.0057,
+      "step": 19877
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6576567078080336e-05,
+      "loss": 0.0022,
+      "step": 19878
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.65655514452715e-05,
+      "loss": 0.0032,
+      "step": 19879
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6554537874050878e-05,
+      "loss": 0.0008,
+      "step": 19880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.654352636460258e-05,
+      "loss": 0.002,
+      "step": 19881
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6532516917110352e-05,
+      "loss": 0.0045,
+      "step": 19882
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6521509531758233e-05,
+      "loss": 0.0031,
+      "step": 19883
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6510504208729942e-05,
+      "loss": 0.0043,
+      "step": 19884
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6499500948209335e-05,
+      "loss": 0.0021,
+      "step": 19885
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6488499750380192e-05,
+      "loss": 0.0005,
+      "step": 19886
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6477500615426206e-05,
+      "loss": 0.0015,
+      "step": 19887
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6466503543531193e-05,
+      "loss": 0.0077,
+      "step": 19888
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6455508534878737e-05,
+      "loss": 0.001,
+      "step": 19889
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.644451558965248e-05,
+      "loss": 0.0056,
+      "step": 19890
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6433524708036043e-05,
+      "loss": 0.0004,
+      "step": 19891
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6422535890213002e-05,
+      "loss": 0.0054,
+      "step": 19892
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6411549136366874e-05,
+      "loss": 0.0037,
+      "step": 19893
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.640056444668115e-05,
+      "loss": 0.0009,
+      "step": 19894
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6389581821339325e-05,
+      "loss": 0.0086,
+      "step": 19895
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.637860126052481e-05,
+      "loss": 0.0027,
+      "step": 19896
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6367622764421053e-05,
+      "loss": 0.0018,
+      "step": 19897
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6356646333211283e-05,
+      "loss": 0.0044,
+      "step": 19898
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.634567196707899e-05,
+      "loss": 0.0053,
+      "step": 19899
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6334699666207325e-05,
+      "loss": 0.0004,
+      "step": 19900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6323729430779662e-05,
+      "loss": 0.002,
+      "step": 19901
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.631276126097915e-05,
+      "loss": 0.0038,
+      "step": 19902
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6301795156989008e-05,
+      "loss": 0.0103,
+      "step": 19903
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6290831118992378e-05,
+      "loss": 0.0006,
+      "step": 19904
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.627986914717238e-05,
+      "loss": 0.0016,
+      "step": 19905
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.626890924171211e-05,
+      "loss": 0.0047,
+      "step": 19906
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6257951402794613e-05,
+      "loss": 0.0018,
+      "step": 19907
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6246995630602908e-05,
+      "loss": 0.0013,
+      "step": 19908
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6236041925319977e-05,
+      "loss": 0.0058,
+      "step": 19909
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6225090287128763e-05,
+      "loss": 0.0027,
+      "step": 19910
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.62141407162122e-05,
+      "loss": 0.0018,
+      "step": 19911
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.620319321275316e-05,
+      "loss": 0.0001,
+      "step": 19912
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6192247776934427e-05,
+      "loss": 0.0021,
+      "step": 19913
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6181304408938925e-05,
+      "loss": 0.0018,
+      "step": 19914
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6170363108949272e-05,
+      "loss": 0.002,
+      "step": 19915
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6159423877148406e-05,
+      "loss": 0.0009,
+      "step": 19916
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6148486713718862e-05,
+      "loss": 0.0003,
+      "step": 19917
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.613755161884335e-05,
+      "loss": 0.0014,
+      "step": 19918
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6126618592704617e-05,
+      "loss": 0.002,
+      "step": 19919
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6115687635485123e-05,
+      "loss": 0.0058,
+      "step": 19920
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6104758747367516e-05,
+      "loss": 0.0003,
+      "step": 19921
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6093831928534287e-05,
+      "loss": 0.0013,
+      "step": 19922
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6082907179167973e-05,
+      "loss": 0.003,
+      "step": 19923
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6071984499450992e-05,
+      "loss": 0.0006,
+      "step": 19924
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6061063889565818e-05,
+      "loss": 0.002,
+      "step": 19925
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.605014534969484e-05,
+      "loss": 0.0002,
+      "step": 19926
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.603922888002039e-05,
+      "loss": 0.0017,
+      "step": 19927
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.602831448072481e-05,
+      "loss": 0.0003,
+      "step": 19928
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.60174021519904e-05,
+      "loss": 0.0013,
+      "step": 19929
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6006491893999438e-05,
+      "loss": 0.0015,
+      "step": 19930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5995583706934053e-05,
+      "loss": 0.0009,
+      "step": 19931
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5984677590976553e-05,
+      "loss": 0.0012,
+      "step": 19932
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.597377354630896e-05,
+      "loss": 0.0,
+      "step": 19933
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5962871573113564e-05,
+      "loss": 0.0002,
+      "step": 19934
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.595197167157228e-05,
+      "loss": 0.0088,
+      "step": 19935
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.594107384186725e-05,
+      "loss": 0.0001,
+      "step": 19936
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.593017808418046e-05,
+      "loss": 0.0024,
+      "step": 19937
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.591928439869388e-05,
+      "loss": 0.0017,
+      "step": 19938
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.590839278558948e-05,
+      "loss": 0.0048,
+      "step": 19939
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5897503245049155e-05,
+      "loss": 0.0004,
+      "step": 19940
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5886615777254788e-05,
+      "loss": 0.0005,
+      "step": 19941
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5875730382388215e-05,
+      "loss": 0.0025,
+      "step": 19942
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5864847060631295e-05,
+      "loss": 0.0008,
+      "step": 19943
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.585396581216566e-05,
+      "loss": 0.0019,
+      "step": 19944
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5843086637173238e-05,
+      "loss": 0.0009,
+      "step": 19945
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.583220953583553e-05,
+      "loss": 0.0022,
+      "step": 19946
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5821334508334358e-05,
+      "loss": 0.0028,
+      "step": 19947
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.581046155485134e-05,
+      "loss": 0.0024,
+      "step": 19948
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5799590675567955e-05,
+      "loss": 0.0039,
+      "step": 19949
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5788721870665936e-05,
+      "loss": 0.0002,
+      "step": 19950
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5777855140326654e-05,
+      "loss": 0.0004,
+      "step": 19951
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.576699048473173e-05,
+      "loss": 0.0012,
+      "step": 19952
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5756127904062557e-05,
+      "loss": 0.0002,
+      "step": 19953
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5745267398500567e-05,
+      "loss": 0.0042,
+      "step": 19954
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5734408968227154e-05,
+      "loss": 0.0049,
+      "step": 19955
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5723552613423687e-05,
+      "loss": 0.0002,
+      "step": 19956
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5712698334271486e-05,
+      "loss": 0.0002,
+      "step": 19957
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.570184613095181e-05,
+      "loss": 0.002,
+      "step": 19958
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.569099600364596e-05,
+      "loss": 0.002,
+      "step": 19959
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5680147952535106e-05,
+      "loss": 0.0002,
+      "step": 19960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.566930197780046e-05,
+      "loss": 0.0037,
+      "step": 19961
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5658458079623157e-05,
+      "loss": 0.0017,
+      "step": 19962
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5647616258184352e-05,
+      "loss": 0.0006,
+      "step": 19963
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5636776513665014e-05,
+      "loss": 0.0033,
+      "step": 19964
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5625938846246337e-05,
+      "loss": 0.0037,
+      "step": 19965
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5615103256109186e-05,
+      "loss": 0.0011,
+      "step": 19966
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5604269743434675e-05,
+      "loss": 0.0015,
+      "step": 19967
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5593438308403617e-05,
+      "loss": 0.0018,
+      "step": 19968
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5582608951196993e-05,
+      "loss": 0.0113,
+      "step": 19969
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.557178167199565e-05,
+      "loss": 0.0031,
+      "step": 19970
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.556095647098043e-05,
+      "loss": 0.0011,
+      "step": 19971
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5550133348332123e-05,
+      "loss": 0.0031,
+      "step": 19972
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5539312304231518e-05,
+      "loss": 0.0001,
+      "step": 19973
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5528493338859334e-05,
+      "loss": 0.0019,
+      "step": 19974
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5517676452396264e-05,
+      "loss": 0.0028,
+      "step": 19975
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5506861645023007e-05,
+      "loss": 0.0011,
+      "step": 19976
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5496048916920076e-05,
+      "loss": 0.0008,
+      "step": 19977
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5485238268268204e-05,
+      "loss": 0.0029,
+      "step": 19978
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5474429699247867e-05,
+      "loss": 0.0004,
+      "step": 19979
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5463623210039635e-05,
+      "loss": 0.0007,
+      "step": 19980
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5452818800823997e-05,
+      "loss": 0.0001,
+      "step": 19981
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5442016471781314e-05,
+      "loss": 0.0051,
+      "step": 19982
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.543121622309215e-05,
+      "loss": 0.0019,
+      "step": 19983
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.542041805493675e-05,
+      "loss": 0.0012,
+      "step": 19984
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5409621967495597e-05,
+      "loss": 0.0026,
+      "step": 19985
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5398827960948886e-05,
+      "loss": 0.0019,
+      "step": 19986
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5388036035476972e-05,
+      "loss": 0.0003,
+      "step": 19987
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5377246191260074e-05,
+      "loss": 0.0002,
+      "step": 19988
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5366458428478403e-05,
+      "loss": 0.0011,
+      "step": 19989
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.535567274731213e-05,
+      "loss": 0.0013,
+      "step": 19990
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5344889147941404e-05,
+      "loss": 0.0005,
+      "step": 19991
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5334107630546357e-05,
+      "loss": 0.0001,
+      "step": 19992
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.532332819530701e-05,
+      "loss": 0.002,
+      "step": 19993
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5312550842403477e-05,
+      "loss": 0.0001,
+      "step": 19994
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5301775572015624e-05,
+      "loss": 0.0059,
+      "step": 19995
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.529100238432359e-05,
+      "loss": 0.0016,
+      "step": 19996
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.528023127950714e-05,
+      "loss": 0.0003,
+      "step": 19997
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5269462257746338e-05,
+      "loss": 0.0003,
+      "step": 19998
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.52586953192209e-05,
+      "loss": 0.0007,
+      "step": 19999
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5247930464110705e-05,
+      "loss": 0.0038,
+      "step": 20000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.523716769259562e-05,
+      "loss": 0.0011,
+      "step": 20001
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.522640700485526e-05,
+      "loss": 0.0014,
+      "step": 20002
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5215648401069522e-05,
+      "loss": 0.0043,
+      "step": 20003
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.520489188141795e-05,
+      "loss": 0.0063,
+      "step": 20004
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5194137446080264e-05,
+      "loss": 0.0016,
+      "step": 20005
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5183385095236057e-05,
+      "loss": 0.0003,
+      "step": 20006
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5172634829064935e-05,
+      "loss": 0.008,
+      "step": 20007
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5161886647746436e-05,
+      "loss": 0.0048,
+      "step": 20008
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5151140551460085e-05,
+      "loss": 0.0068,
+      "step": 20009
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5140396540385353e-05,
+      "loss": 0.0079,
+      "step": 20010
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5129654614701685e-05,
+      "loss": 0.0002,
+      "step": 20011
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5118914774588495e-05,
+      "loss": 0.0037,
+      "step": 20012
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5108177020225163e-05,
+      "loss": 0.0006,
+      "step": 20013
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.509744135179106e-05,
+      "loss": 0.0065,
+      "step": 20014
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5086707769465374e-05,
+      "loss": 0.0001,
+      "step": 20015
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5075976273427534e-05,
+      "loss": 0.0001,
+      "step": 20016
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5065246863856642e-05,
+      "loss": 0.0026,
+      "step": 20017
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5054519540932022e-05,
+      "loss": 0.0004,
+      "step": 20018
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.504379430483273e-05,
+      "loss": 0.0001,
+      "step": 20019
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.503307115573796e-05,
+      "loss": 0.0001,
+      "step": 20020
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.502235009382678e-05,
+      "loss": 0.0008,
+      "step": 20021
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5011631119278286e-05,
+      "loss": 0.002,
+      "step": 20022
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5000914232271462e-05,
+      "loss": 0.0026,
+      "step": 20023
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4990199432985337e-05,
+      "loss": 0.0015,
+      "step": 20024
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4979486721598845e-05,
+      "loss": 0.0023,
+      "step": 20025
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.496877609829093e-05,
+      "loss": 0.0027,
+      "step": 20026
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.49580675632405e-05,
+      "loss": 0.0038,
+      "step": 20027
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4947361116626297e-05,
+      "loss": 0.0032,
+      "step": 20028
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.493665675862729e-05,
+      "loss": 0.0057,
+      "step": 20029
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4925954489422117e-05,
+      "loss": 0.0008,
+      "step": 20030
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.491525430918964e-05,
+      "loss": 0.0007,
+      "step": 20031
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4904556218108567e-05,
+      "loss": 0.0002,
+      "step": 20032
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4893860216357452e-05,
+      "loss": 0.0034,
+      "step": 20033
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.488316630411512e-05,
+      "loss": 0.0011,
+      "step": 20034
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4872474481559994e-05,
+      "loss": 0.0049,
+      "step": 20035
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4861784748870817e-05,
+      "loss": 0.0019,
+      "step": 20036
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4851097106226024e-05,
+      "loss": 0.0062,
+      "step": 20037
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.484041155380413e-05,
+      "loss": 0.0002,
+      "step": 20038
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.482972809178362e-05,
+      "loss": 0.0018,
+      "step": 20039
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4819046720342934e-05,
+      "loss": 0.0009,
+      "step": 20040
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4808367439660465e-05,
+      "loss": 0.0001,
+      "step": 20041
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.479769024991457e-05,
+      "loss": 0.0005,
+      "step": 20042
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.478701515128359e-05,
+      "loss": 0.0007,
+      "step": 20043
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.47763421439458e-05,
+      "loss": 0.002,
+      "step": 20044
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.476567122807952e-05,
+      "loss": 0.001,
+      "step": 20045
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.475500240386286e-05,
+      "loss": 0.0003,
+      "step": 20046
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.474433567147413e-05,
+      "loss": 0.0007,
+      "step": 20047
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4733671031091368e-05,
+      "loss": 0.0015,
+      "step": 20048
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.472300848289283e-05,
+      "loss": 0.0002,
+      "step": 20049
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4712348027056494e-05,
+      "loss": 0.003,
+      "step": 20050
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.470168966376045e-05,
+      "loss": 0.0024,
+      "step": 20051
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.46910333931827e-05,
+      "loss": 0.009,
+      "step": 20052
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4680379215501178e-05,
+      "loss": 0.0001,
+      "step": 20053
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.466972713089397e-05,
+      "loss": 0.0016,
+      "step": 20054
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4659077139538862e-05,
+      "loss": 0.0044,
+      "step": 20055
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4648429241613755e-05,
+      "loss": 0.0026,
+      "step": 20056
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.46377834372965e-05,
+      "loss": 0.0007,
+      "step": 20057
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4627139726764904e-05,
+      "loss": 0.0024,
+      "step": 20058
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4616498110196732e-05,
+      "loss": 0.0054,
+      "step": 20059
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4605858587769717e-05,
+      "loss": 0.0005,
+      "step": 20060
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4595221159661555e-05,
+      "loss": 0.0008,
+      "step": 20061
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.458458582604992e-05,
+      "loss": 0.0033,
+      "step": 20062
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.457395258711245e-05,
+      "loss": 0.0012,
+      "step": 20063
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4563321443026722e-05,
+      "loss": 0.0024,
+      "step": 20064
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4552692393970346e-05,
+      "loss": 0.0015,
+      "step": 20065
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.454206544012073e-05,
+      "loss": 0.0003,
+      "step": 20066
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.453144058165551e-05,
+      "loss": 0.0009,
+      "step": 20067
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4520817818752032e-05,
+      "loss": 0.0073,
+      "step": 20068
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4510197151587767e-05,
+      "loss": 0.0019,
+      "step": 20069
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.449957858034008e-05,
+      "loss": 0.0027,
+      "step": 20070
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.448896210518632e-05,
+      "loss": 0.0061,
+      "step": 20071
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4478347726303838e-05,
+      "loss": 0.0015,
+      "step": 20072
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4467735443869868e-05,
+      "loss": 0.0041,
+      "step": 20073
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.445712525806167e-05,
+      "loss": 0.0003,
+      "step": 20074
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4446517169056474e-05,
+      "loss": 0.0017,
+      "step": 20075
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4435911177031432e-05,
+      "loss": 0.0037,
+      "step": 20076
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4425307282163704e-05,
+      "loss": 0.0152,
+      "step": 20077
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4414705484630398e-05,
+      "loss": 0.0002,
+      "step": 20078
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4404105784608523e-05,
+      "loss": 0.0008,
+      "step": 20079
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4393508182275207e-05,
+      "loss": 0.0053,
+      "step": 20080
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4382912677807355e-05,
+      "loss": 0.0015,
+      "step": 20081
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.437231927138204e-05,
+      "loss": 0.0005,
+      "step": 20082
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.436172796317609e-05,
+      "loss": 0.0008,
+      "step": 20083
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4351138753366416e-05,
+      "loss": 0.0112,
+      "step": 20084
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4340551642129973e-05,
+      "loss": 0.002,
+      "step": 20085
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4329966629643455e-05,
+      "loss": 0.0022,
+      "step": 20086
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4319383716083768e-05,
+      "loss": 0.0009,
+      "step": 20087
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4308802901627587e-05,
+      "loss": 0.0037,
+      "step": 20088
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4298224186451643e-05,
+      "loss": 0.0019,
+      "step": 20089
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4287647570732656e-05,
+      "loss": 0.0006,
+      "step": 20090
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4277073054647233e-05,
+      "loss": 0.0002,
+      "step": 20091
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.426650063837202e-05,
+      "loss": 0.0002,
+      "step": 20092
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4255930322083578e-05,
+      "loss": 0.0017,
+      "step": 20093
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.424536210595846e-05,
+      "loss": 0.0026,
+      "step": 20094
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4234795990173177e-05,
+      "loss": 0.0094,
+      "step": 20095
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4224231974904242e-05,
+      "loss": 0.0003,
+      "step": 20096
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4213670060327973e-05,
+      "loss": 0.0001,
+      "step": 20097
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4203110246620916e-05,
+      "loss": 0.0035,
+      "step": 20098
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4192552533959316e-05,
+      "loss": 0.0041,
+      "step": 20099
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.418199692251964e-05,
+      "loss": 0.0029,
+      "step": 20100
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4171443412478074e-05,
+      "loss": 0.0048,
+      "step": 20101
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4160892004010924e-05,
+      "loss": 0.002,
+      "step": 20102
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4150342697294412e-05,
+      "loss": 0.0007,
+      "step": 20103
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.413979549250473e-05,
+      "loss": 0.0021,
+      "step": 20104
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4129250389818043e-05,
+      "loss": 0.0023,
+      "step": 20105
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4118707389410474e-05,
+      "loss": 0.0073,
+      "step": 20106
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4108166491458116e-05,
+      "loss": 0.0027,
+      "step": 20107
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4097627696137008e-05,
+      "loss": 0.0049,
+      "step": 20108
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4087091003623177e-05,
+      "loss": 0.0006,
+      "step": 20109
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.40765564140926e-05,
+      "loss": 0.0065,
+      "step": 20110
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4066023927721275e-05,
+      "loss": 0.0038,
+      "step": 20111
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.4055493544684985e-05,
+      "loss": 0.0042,
+      "step": 20112
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.4044965265159726e-05,
+      "loss": 0.0002,
+      "step": 20113
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.403443908932131e-05,
+      "loss": 0.0036,
+      "step": 20114
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.402391501734553e-05,
+      "loss": 0.0003,
+      "step": 20115
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.4013393049408208e-05,
+      "loss": 0.0005,
+      "step": 20116
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.4002873185684973e-05,
+      "loss": 0.0019,
+      "step": 20117
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.399235542635165e-05,
+      "loss": 0.0049,
+      "step": 20118
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3981839771583834e-05,
+      "loss": 0.0003,
+      "step": 20119
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3971326221557147e-05,
+      "loss": 0.0003,
+      "step": 20120
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.396081477644722e-05,
+      "loss": 0.0009,
+      "step": 20121
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.395030543642959e-05,
+      "loss": 0.0013,
+      "step": 20122
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3939798201679804e-05,
+      "loss": 0.0001,
+      "step": 20123
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.392929307237334e-05,
+      "loss": 0.0015,
+      "step": 20124
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.391879004868567e-05,
+      "loss": 0.0007,
+      "step": 20125
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.390828913079219e-05,
+      "loss": 0.0029,
+      "step": 20126
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3897790318868293e-05,
+      "loss": 0.0011,
+      "step": 20127
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3887293613089337e-05,
+      "loss": 0.0,
+      "step": 20128
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3876799013630666e-05,
+      "loss": 0.0014,
+      "step": 20129
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3866306520667455e-05,
+      "loss": 0.0015,
+      "step": 20130
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.38558161343751e-05,
+      "loss": 0.0001,
+      "step": 20131
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3845327854928644e-05,
+      "loss": 0.0015,
+      "step": 20132
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3834841682503424e-05,
+      "loss": 0.0004,
+      "step": 20133
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3824357617274475e-05,
+      "loss": 0.0002,
+      "step": 20134
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.381387565941692e-05,
+      "loss": 0.0019,
+      "step": 20135
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.380339580910582e-05,
+      "loss": 0.0001,
+      "step": 20136
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.37929180665162e-05,
+      "loss": 0.0039,
+      "step": 20137
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3782442431823152e-05,
+      "loss": 0.0013,
+      "step": 20138
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3771968905201537e-05,
+      "loss": 0.0002,
+      "step": 20139
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.376149748682632e-05,
+      "loss": 0.001,
+      "step": 20140
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3751028176872372e-05,
+      "loss": 0.001,
+      "step": 20141
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3740560975514606e-05,
+      "loss": 0.002,
+      "step": 20142
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.373009588292773e-05,
+      "loss": 0.0004,
+      "step": 20143
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.371963289928664e-05,
+      "loss": 0.0002,
+      "step": 20144
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3709172024766044e-05,
+      "loss": 0.0018,
+      "step": 20145
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3698713259540686e-05,
+      "loss": 0.0024,
+      "step": 20146
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3688256603785245e-05,
+      "loss": 0.0001,
+      "step": 20147
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3677802057674295e-05,
+      "loss": 0.0001,
+      "step": 20148
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3667349621382547e-05,
+      "loss": 0.0048,
+      "step": 20149
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.365689929508448e-05,
+      "loss": 0.0045,
+      "step": 20150
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3646451078954754e-05,
+      "loss": 0.0007,
+      "step": 20151
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.363600497316776e-05,
+      "loss": 0.0022,
+      "step": 20152
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3625560977898013e-05,
+      "loss": 0.0023,
+      "step": 20153
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.361511909331995e-05,
+      "loss": 0.009,
+      "step": 20154
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3604679319607954e-05,
+      "loss": 0.0063,
+      "step": 20155
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.359424165693642e-05,
+      "loss": 0.002,
+      "step": 20156
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3583806105479653e-05,
+      "loss": 0.0027,
+      "step": 20157
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.357337266541195e-05,
+      "loss": 0.0032,
+      "step": 20158
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.356294133690757e-05,
+      "loss": 0.0024,
+      "step": 20159
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3552512120140725e-05,
+      "loss": 0.0001,
+      "step": 20160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3542085015285626e-05,
+      "loss": 0.0057,
+      "step": 20161
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.353166002251645e-05,
+      "loss": 0.0048,
+      "step": 20162
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.352123714200721e-05,
+      "loss": 0.0045,
+      "step": 20163
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.351081637393213e-05,
+      "loss": 0.0036,
+      "step": 20164
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3500397718465087e-05,
+      "loss": 0.0049,
+      "step": 20165
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3489981175780243e-05,
+      "loss": 0.0002,
+      "step": 20166
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.347956674605154e-05,
+      "loss": 0.0045,
+      "step": 20167
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3469154429452846e-05,
+      "loss": 0.0002,
+      "step": 20168
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3458744226158166e-05,
+      "loss": 0.0001,
+      "step": 20169
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3448336136341295e-05,
+      "loss": 0.0001,
+      "step": 20170
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3437930160176095e-05,
+      "loss": 0.0003,
+      "step": 20171
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3427526297836374e-05,
+      "loss": 0.0006,
+      "step": 20172
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3417124549495862e-05,
+      "loss": 0.0137,
+      "step": 20173
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.340672491532834e-05,
+      "loss": 0.0035,
+      "step": 20174
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3396327395507448e-05,
+      "loss": 0.0016,
+      "step": 20175
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.338593199020688e-05,
+      "loss": 0.0019,
+      "step": 20176
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.337553869960025e-05,
+      "loss": 0.0021,
+      "step": 20177
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.336514752386115e-05,
+      "loss": 0.0033,
+      "step": 20178
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3354758463163108e-05,
+      "loss": 0.0002,
+      "step": 20179
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.334437151767972e-05,
+      "loss": 0.0044,
+      "step": 20180
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3333986687584317e-05,
+      "loss": 0.0063,
+      "step": 20181
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.332360397305052e-05,
+      "loss": 0.0017,
+      "step": 20182
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3313223374251576e-05,
+      "loss": 0.0014,
+      "step": 20183
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3302844891361018e-05,
+      "loss": 0.0,
+      "step": 20184
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3292468524552048e-05,
+      "loss": 0.001,
+      "step": 20185
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3282094273998054e-05,
+      "loss": 0.0023,
+      "step": 20186
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.327172213987229e-05,
+      "loss": 0.008,
+      "step": 20187
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.326135212234797e-05,
+      "loss": 0.0015,
+      "step": 20188
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3250984221598308e-05,
+      "loss": 0.0013,
+      "step": 20189
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.324061843779646e-05,
+      "loss": 0.0006,
+      "step": 20190
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3230254771115557e-05,
+      "loss": 0.0051,
+      "step": 20191
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3219893221728697e-05,
+      "loss": 0.0007,
+      "step": 20192
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3209533789808986e-05,
+      "loss": 0.0023,
+      "step": 20193
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.319917647552931e-05,
+      "loss": 0.0006,
+      "step": 20194
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3188821279062802e-05,
+      "loss": 0.0017,
+      "step": 20195
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3178468200582302e-05,
+      "loss": 0.0018,
+      "step": 20196
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3168117240260796e-05,
+      "loss": 0.0055,
+      "step": 20197
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3157768398271188e-05,
+      "loss": 0.0002,
+      "step": 20198
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3147421674786197e-05,
+      "loss": 0.0005,
+      "step": 20199
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3137077069978792e-05,
+      "loss": 0.0001,
+      "step": 20200
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3126734584021617e-05,
+      "loss": 0.0045,
+      "step": 20201
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3116394217087515e-05,
+      "loss": 0.0049,
+      "step": 20202
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3106055969349112e-05,
+      "loss": 0.0018,
+      "step": 20203
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3095719840979093e-05,
+      "loss": 0.0007,
+      "step": 20204
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3085385832150117e-05,
+      "loss": 0.0012,
+      "step": 20205
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.307505394303476e-05,
+      "loss": 0.0009,
+      "step": 20206
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.306472417380559e-05,
+      "loss": 0.0005,
+      "step": 20207
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3054396524635128e-05,
+      "loss": 0.0003,
+      "step": 20208
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3044070995695872e-05,
+      "loss": 0.0002,
+      "step": 20209
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.3033747587160272e-05,
+      "loss": 0.0054,
+      "step": 20210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.302342629920077e-05,
+      "loss": 0.0006,
+      "step": 20211
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.301310713198971e-05,
+      "loss": 0.003,
+      "step": 20212
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.3002790085699523e-05,
+      "loss": 0.0002,
+      "step": 20213
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2992475160502388e-05,
+      "loss": 0.003,
+      "step": 20214
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.298216235657073e-05,
+      "loss": 0.0017,
+      "step": 20215
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2971851674076685e-05,
+      "loss": 0.0033,
+      "step": 20216
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.296154311319251e-05,
+      "loss": 0.0006,
+      "step": 20217
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2951236674090368e-05,
+      "loss": 0.0004,
+      "step": 20218
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2940932356942354e-05,
+      "loss": 0.0008,
+      "step": 20219
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.293063016192068e-05,
+      "loss": 0.0025,
+      "step": 20220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2920330089197296e-05,
+      "loss": 0.0039,
+      "step": 20221
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2910032138944296e-05,
+      "loss": 0.0005,
+      "step": 20222
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2899736311333653e-05,
+      "loss": 0.0026,
+      "step": 20223
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.288944260653734e-05,
+      "loss": 0.0002,
+      "step": 20224
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.287915102472726e-05,
+      "loss": 0.002,
+      "step": 20225
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2868861566075365e-05,
+      "loss": 0.0013,
+      "step": 20226
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2858574230753378e-05,
+      "loss": 0.0026,
+      "step": 20227
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2848289018933236e-05,
+      "loss": 0.0017,
+      "step": 20228
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2838005930786678e-05,
+      "loss": 0.0012,
+      "step": 20229
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.282772496648547e-05,
+      "loss": 0.0003,
+      "step": 20230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2817446126201345e-05,
+      "loss": 0.0004,
+      "step": 20231
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2807169410105858e-05,
+      "loss": 0.003,
+      "step": 20232
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2796894818370808e-05,
+      "loss": 0.0001,
+      "step": 20233
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.278662235116766e-05,
+      "loss": 0.01,
+      "step": 20234
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2776352008668125e-05,
+      "loss": 0.0023,
+      "step": 20235
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.276608379104363e-05,
+      "loss": 0.0007,
+      "step": 20236
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2755817698465713e-05,
+      "loss": 0.0031,
+      "step": 20237
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.274555373110581e-05,
+      "loss": 0.0024,
+      "step": 20238
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.273529188913539e-05,
+      "loss": 0.0015,
+      "step": 20239
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.272503217272581e-05,
+      "loss": 0.002,
+      "step": 20240
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.271477458204844e-05,
+      "loss": 0.0025,
+      "step": 20241
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2704519117274606e-05,
+      "loss": 0.0005,
+      "step": 20242
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2694265778575604e-05,
+      "loss": 0.0026,
+      "step": 20243
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2684014566122683e-05,
+      "loss": 0.0001,
+      "step": 20244
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2673765480086986e-05,
+      "loss": 0.0005,
+      "step": 20245
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.266351852063983e-05,
+      "loss": 0.0058,
+      "step": 20246
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.265327368795219e-05,
+      "loss": 0.0012,
+      "step": 20247
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.264303098219535e-05,
+      "loss": 0.0006,
+      "step": 20248
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2632790403540252e-05,
+      "loss": 0.0011,
+      "step": 20249
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2622551952157947e-05,
+      "loss": 0.0005,
+      "step": 20250
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.261231562821953e-05,
+      "loss": 0.0002,
+      "step": 20251
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.260208143189583e-05,
+      "loss": 0.0009,
+      "step": 20252
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2591849363357927e-05,
+      "loss": 0.0001,
+      "step": 20253
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2581619422776603e-05,
+      "loss": 0.0003,
+      "step": 20254
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.257139161032277e-05,
+      "loss": 0.0037,
+      "step": 20255
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2561165926167208e-05,
+      "loss": 0.001,
+      "step": 20256
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2550942370480745e-05,
+      "loss": 0.0044,
+      "step": 20257
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2540720943434133e-05,
+      "loss": 0.0014,
+      "step": 20258
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2530501645198064e-05,
+      "loss": 0.0027,
+      "step": 20259
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2520284475943234e-05,
+      "loss": 0.0014,
+      "step": 20260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2510069435840293e-05,
+      "loss": 0.0005,
+      "step": 20261
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2499856525059838e-05,
+      "loss": 0.0004,
+      "step": 20262
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2489645743772467e-05,
+      "loss": 0.0005,
+      "step": 20263
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2479437092148723e-05,
+      "loss": 0.0007,
+      "step": 20264
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2469230570359037e-05,
+      "loss": 0.0014,
+      "step": 20265
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.245902617857399e-05,
+      "loss": 0.0062,
+      "step": 20266
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.244882391696393e-05,
+      "loss": 0.0013,
+      "step": 20267
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2438623785699273e-05,
+      "loss": 0.001,
+      "step": 20268
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2428425784950394e-05,
+      "loss": 0.0031,
+      "step": 20269
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.241822991488761e-05,
+      "loss": 0.0002,
+      "step": 20270
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2408036175681214e-05,
+      "loss": 0.0002,
+      "step": 20271
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2397844567501478e-05,
+      "loss": 0.0053,
+      "step": 20272
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2387655090518575e-05,
+      "loss": 0.0006,
+      "step": 20273
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.237746774490274e-05,
+      "loss": 0.0013,
+      "step": 20274
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2367282530824082e-05,
+      "loss": 0.0075,
+      "step": 20275
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2357099448452738e-05,
+      "loss": 0.0022,
+      "step": 20276
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2346918497958815e-05,
+      "loss": 0.0042,
+      "step": 20277
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2336739679512232e-05,
+      "loss": 0.0007,
+      "step": 20278
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2326562993283154e-05,
+      "loss": 0.0005,
+      "step": 20279
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2316388439441407e-05,
+      "loss": 0.0001,
+      "step": 20280
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2306216018157026e-05,
+      "loss": 0.0021,
+      "step": 20281
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.229604572959992e-05,
+      "loss": 0.0032,
+      "step": 20282
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2285877573939825e-05,
+      "loss": 0.0015,
+      "step": 20283
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2275711551346736e-05,
+      "loss": 0.0007,
+      "step": 20284
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2265547661990286e-05,
+      "loss": 0.004,
+      "step": 20285
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.225538590604037e-05,
+      "loss": 0.0018,
+      "step": 20286
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.22452262836666e-05,
+      "loss": 0.003,
+      "step": 20287
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.223506879503873e-05,
+      "loss": 0.0004,
+      "step": 20288
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.222491344032637e-05,
+      "loss": 0.002,
+      "step": 20289
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.221476021969915e-05,
+      "loss": 0.0037,
+      "step": 20290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.220460913332664e-05,
+      "loss": 0.0002,
+      "step": 20291
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2194460181378397e-05,
+      "loss": 0.0003,
+      "step": 20292
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2184313364023925e-05,
+      "loss": 0.0002,
+      "step": 20293
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.217416868143267e-05,
+      "loss": 0.0013,
+      "step": 20294
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2164026133774144e-05,
+      "loss": 0.0004,
+      "step": 20295
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2153885721217594e-05,
+      "loss": 0.0007,
+      "step": 20296
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2143747443932574e-05,
+      "loss": 0.0016,
+      "step": 20297
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2133611302088227e-05,
+      "loss": 0.0015,
+      "step": 20298
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.212347729585402e-05,
+      "loss": 0.0013,
+      "step": 20299
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2113345425399083e-05,
+      "loss": 0.0011,
+      "step": 20300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.210321569089268e-05,
+      "loss": 0.0001,
+      "step": 20301
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2093088092504008e-05,
+      "loss": 0.0149,
+      "step": 20302
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2082962630402168e-05,
+      "loss": 0.0085,
+      "step": 20303
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2072839304756373e-05,
+      "loss": 0.0016,
+      "step": 20304
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2062718115735604e-05,
+      "loss": 0.0038,
+      "step": 20305
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2052599063508962e-05,
+      "loss": 0.0003,
+      "step": 20306
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.204248214824541e-05,
+      "loss": 0.0003,
+      "step": 20307
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2032367370113963e-05,
+      "loss": 0.0003,
+      "step": 20308
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2022254729283518e-05,
+      "loss": 0.0028,
+      "step": 20309
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2012144225923005e-05,
+      "loss": 0.0021,
+      "step": 20310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2002035860201277e-05,
+      "loss": 0.0008,
+      "step": 20311
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1991929632287166e-05,
+      "loss": 0.0006,
+      "step": 20312
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1981825542349462e-05,
+      "loss": 0.0014,
+      "step": 20313
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1971723590556922e-05,
+      "loss": 0.0027,
+      "step": 20314
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.196162377707829e-05,
+      "loss": 0.001,
+      "step": 20315
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1951526102082184e-05,
+      "loss": 0.0038,
+      "step": 20316
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.194143056573735e-05,
+      "loss": 0.0089,
+      "step": 20317
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1931337168212336e-05,
+      "loss": 0.001,
+      "step": 20318
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.192124590967573e-05,
+      "loss": 0.0002,
+      "step": 20319
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1911156790296086e-05,
+      "loss": 0.003,
+      "step": 20320
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1901069810241896e-05,
+      "loss": 0.0051,
+      "step": 20321
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.189098496968165e-05,
+      "loss": 0.0028,
+      "step": 20322
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.188090226878379e-05,
+      "loss": 0.0004,
+      "step": 20323
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.187082170771669e-05,
+      "loss": 0.0001,
+      "step": 20324
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1860743286648736e-05,
+      "loss": 0.0013,
+      "step": 20325
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.185066700574824e-05,
+      "loss": 0.0007,
+      "step": 20326
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1840592865183522e-05,
+      "loss": 0.0011,
+      "step": 20327
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1830520865122842e-05,
+      "loss": 0.0019,
+      "step": 20328
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1820451005734337e-05,
+      "loss": 0.0043,
+      "step": 20329
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1810383287186318e-05,
+      "loss": 0.0005,
+      "step": 20330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1800317709646804e-05,
+      "loss": 0.0004,
+      "step": 20331
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.179025427328406e-05,
+      "loss": 0.0001,
+      "step": 20332
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1780192978266032e-05,
+      "loss": 0.0043,
+      "step": 20333
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.177013382476079e-05,
+      "loss": 0.0041,
+      "step": 20334
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.176007681293643e-05,
+      "loss": 0.0019,
+      "step": 20335
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.175002194296078e-05,
+      "loss": 0.0003,
+      "step": 20336
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1739969215001933e-05,
+      "loss": 0.0026,
+      "step": 20337
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1729918629227667e-05,
+      "loss": 0.0013,
+      "step": 20338
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1719870185805892e-05,
+      "loss": 0.0113,
+      "step": 20339
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.170982388490443e-05,
+      "loss": 0.0007,
+      "step": 20340
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1699779726691064e-05,
+      "loss": 0.0001,
+      "step": 20341
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.168973771133357e-05,
+      "loss": 0.0028,
+      "step": 20342
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1679697838999653e-05,
+      "loss": 0.0002,
+      "step": 20343
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1669660109857013e-05,
+      "loss": 0.0002,
+      "step": 20344
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1659624524073295e-05,
+      "loss": 0.0002,
+      "step": 20345
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.164959108181613e-05,
+      "loss": 0.0035,
+      "step": 20346
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1639559783253006e-05,
+      "loss": 0.0003,
+      "step": 20347
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.162953062855159e-05,
+      "loss": 0.0001,
+      "step": 20348
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1619503617879276e-05,
+      "loss": 0.0002,
+      "step": 20349
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1609478751403646e-05,
+      "loss": 0.0035,
+      "step": 20350
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.159945602929204e-05,
+      "loss": 0.0017,
+      "step": 20351
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.15894354517119e-05,
+      "loss": 0.0019,
+      "step": 20352
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.157941701883058e-05,
+      "loss": 0.0039,
+      "step": 20353
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1569400730815396e-05,
+      "loss": 0.0015,
+      "step": 20354
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.155938658783366e-05,
+      "loss": 0.0072,
+      "step": 20355
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.154937459005261e-05,
+      "loss": 0.0014,
+      "step": 20356
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1539364737639476e-05,
+      "loss": 0.0006,
+      "step": 20357
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.152935703076144e-05,
+      "loss": 0.0007,
+      "step": 20358
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.151935146958564e-05,
+      "loss": 0.0011,
+      "step": 20359
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.15093480542792e-05,
+      "loss": 0.0028,
+      "step": 20360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.149934678500923e-05,
+      "loss": 0.0017,
+      "step": 20361
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1489347661942658e-05,
+      "loss": 0.0001,
+      "step": 20362
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.147935068524658e-05,
+      "loss": 0.0011,
+      "step": 20363
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1469355855088005e-05,
+      "loss": 0.0003,
+      "step": 20364
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1459363171633726e-05,
+      "loss": 0.0012,
+      "step": 20365
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1449372635050777e-05,
+      "loss": 0.0012,
+      "step": 20366
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.143938424550591e-05,
+      "loss": 0.0024,
+      "step": 20367
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1429398003166053e-05,
+      "loss": 0.0002,
+      "step": 20368
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.141941390819793e-05,
+      "loss": 0.002,
+      "step": 20369
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1409431960768303e-05,
+      "loss": 0.0004,
+      "step": 20370
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1399452161043873e-05,
+      "loss": 0.0008,
+      "step": 20371
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1389474509191374e-05,
+      "loss": 0.0084,
+      "step": 20372
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1379499005377404e-05,
+      "loss": 0.0018,
+      "step": 20373
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.136952564976858e-05,
+      "loss": 0.002,
+      "step": 20374
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.135955444253149e-05,
+      "loss": 0.0005,
+      "step": 20375
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.134958538383268e-05,
+      "loss": 0.0002,
+      "step": 20376
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.133961847383863e-05,
+      "loss": 0.009,
+      "step": 20377
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1329653712715817e-05,
+      "loss": 0.0025,
+      "step": 20378
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.13196911006307e-05,
+      "loss": 0.0021,
+      "step": 20379
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1309730637749577e-05,
+      "loss": 0.0032,
+      "step": 20380
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.129977232423894e-05,
+      "loss": 0.0009,
+      "step": 20381
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1289816160264984e-05,
+      "loss": 0.0001,
+      "step": 20382
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.127986214599411e-05,
+      "loss": 0.0002,
+      "step": 20383
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1269910281592483e-05,
+      "loss": 0.0068,
+      "step": 20384
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1259960567226357e-05,
+      "loss": 0.0036,
+      "step": 20385
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.12500130030619e-05,
+      "loss": 0.0007,
+      "step": 20386
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1240067589265226e-05,
+      "loss": 0.0038,
+      "step": 20387
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1230124326002535e-05,
+      "loss": 0.0023,
+      "step": 20388
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.122018321343981e-05,
+      "loss": 0.0008,
+      "step": 20389
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1210244251743124e-05,
+      "loss": 0.0028,
+      "step": 20390
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.120030744107846e-05,
+      "loss": 0.0014,
+      "step": 20391
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1190372781611825e-05,
+      "loss": 0.0075,
+      "step": 20392
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.118044027350903e-05,
+      "loss": 0.0004,
+      "step": 20393
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.117050991693609e-05,
+      "loss": 0.0001,
+      "step": 20394
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1160581712058823e-05,
+      "loss": 0.0079,
+      "step": 20395
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1150655659043025e-05,
+      "loss": 0.0046,
+      "step": 20396
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1140731758054552e-05,
+      "loss": 0.0002,
+      "step": 20397
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.113081000925902e-05,
+      "loss": 0.0002,
+      "step": 20398
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.112089041282228e-05,
+      "loss": 0.0023,
+      "step": 20399
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1110972968909877e-05,
+      "loss": 0.0042,
+      "step": 20400
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1101057677687604e-05,
+      "loss": 0.0002,
+      "step": 20401
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1091144539320944e-05,
+      "loss": 0.0005,
+      "step": 20402
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.108123355397549e-05,
+      "loss": 0.005,
+      "step": 20403
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.10713247218168e-05,
+      "loss": 0.0024,
+      "step": 20404
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.106141804301036e-05,
+      "loss": 0.0009,
+      "step": 20405
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1051513517721613e-05,
+      "loss": 0.0026,
+      "step": 20406
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1041611146116018e-05,
+      "loss": 0.0009,
+      "step": 20407
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.103171092835892e-05,
+      "loss": 0.0005,
+      "step": 20408
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1021812864615707e-05,
+      "loss": 0.0025,
+      "step": 20409
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1011916955051693e-05,
+      "loss": 0.0057,
+      "step": 20410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1002023199832134e-05,
+      "loss": 0.0007,
+      "step": 20411
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0992131599122314e-05,
+      "loss": 0.003,
+      "step": 20412
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0982242153087363e-05,
+      "loss": 0.0005,
+      "step": 20413
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0972354861892572e-05,
+      "loss": 0.0035,
+      "step": 20414
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0962469725702985e-05,
+      "loss": 0.0035,
+      "step": 20415
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0952586744683674e-05,
+      "loss": 0.0047,
+      "step": 20416
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.094270591899984e-05,
+      "loss": 0.0002,
+      "step": 20417
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.093282724881637e-05,
+      "loss": 0.0023,
+      "step": 20418
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0922950734298377e-05,
+      "loss": 0.0008,
+      "step": 20419
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0913076375610715e-05,
+      "loss": 0.001,
+      "step": 20420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0903204172918352e-05,
+      "loss": 0.003,
+      "step": 20421
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0893334126386174e-05,
+      "loss": 0.0023,
+      "step": 20422
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0883466236179014e-05,
+      "loss": 0.0003,
+      "step": 20423
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.087360050246169e-05,
+      "loss": 0.0004,
+      "step": 20424
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.086373692539899e-05,
+      "loss": 0.0016,
+      "step": 20425
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0853875505155642e-05,
+      "loss": 0.0022,
+      "step": 20426
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.084401624189634e-05,
+      "loss": 0.0041,
+      "step": 20427
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0834159135785777e-05,
+      "loss": 0.0053,
+      "step": 20428
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.082430418698858e-05,
+      "loss": 0.0001,
+      "step": 20429
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0814451395669372e-05,
+      "loss": 0.0016,
+      "step": 20430
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0804600761992602e-05,
+      "loss": 0.0018,
+      "step": 20431
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.079475228612296e-05,
+      "loss": 0.0002,
+      "step": 20432
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0784905968224756e-05,
+      "loss": 0.0005,
+      "step": 20433
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.077506180846262e-05,
+      "loss": 0.0007,
+      "step": 20434
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0765219807000843e-05,
+      "loss": 0.0001,
+      "step": 20435
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0755379964003837e-05,
+      "loss": 0.0027,
+      "step": 20436
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.074554227963596e-05,
+      "loss": 0.0037,
+      "step": 20437
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0735706754061505e-05,
+      "loss": 0.0054,
+      "step": 20438
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0725873387444746e-05,
+      "loss": 0.0103,
+      "step": 20439
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0716042179949938e-05,
+      "loss": 0.0003,
+      "step": 20440
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0706213131741244e-05,
+      "loss": 0.0003,
+      "step": 20441
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0696386242982872e-05,
+      "loss": 0.002,
+      "step": 20442
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.068656151383895e-05,
+      "loss": 0.0002,
+      "step": 20443
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0676738944473485e-05,
+      "loss": 0.0059,
+      "step": 20444
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0666918535050665e-05,
+      "loss": 0.0012,
+      "step": 20445
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0657100285734367e-05,
+      "loss": 0.0026,
+      "step": 20446
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0647284196688685e-05,
+      "loss": 0.004,
+      "step": 20447
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.063747026807755e-05,
+      "loss": 0.0014,
+      "step": 20448
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0627658500064797e-05,
+      "loss": 0.0005,
+      "step": 20449
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0617848892814416e-05,
+      "loss": 0.0001,
+      "step": 20450
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0608041446490103e-05,
+      "loss": 0.0035,
+      "step": 20451
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0598236161255833e-05,
+      "loss": 0.0007,
+      "step": 20452
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0588433037275238e-05,
+      "loss": 0.0035,
+      "step": 20453
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0578632074712088e-05,
+      "loss": 0.0033,
+      "step": 20454
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0568833273730083e-05,
+      "loss": 0.0029,
+      "step": 20455
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.05590366344929e-05,
+      "loss": 0.0001,
+      "step": 20456
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0549242157164115e-05,
+      "loss": 0.002,
+      "step": 20457
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0539449841907367e-05,
+      "loss": 0.0037,
+      "step": 20458
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0529659688886162e-05,
+      "loss": 0.0009,
+      "step": 20459
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0519871698264035e-05,
+      "loss": 0.0022,
+      "step": 20460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0510085870204466e-05,
+      "loss": 0.0027,
+      "step": 20461
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.05003022048709e-05,
+      "loss": 0.0006,
+      "step": 20462
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.049052070242675e-05,
+      "loss": 0.0024,
+      "step": 20463
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.048074136303532e-05,
+      "loss": 0.0015,
+      "step": 20464
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.047096418686006e-05,
+      "loss": 0.0016,
+      "step": 20465
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0461189174064163e-05,
+      "loss": 0.0042,
+      "step": 20466
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0451416324810927e-05,
+      "loss": 0.001,
+      "step": 20467
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0441645639263586e-05,
+      "loss": 0.0003,
+      "step": 20468
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0431877117585284e-05,
+      "loss": 0.0005,
+      "step": 20469
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.042211075993928e-05,
+      "loss": 0.0001,
+      "step": 20470
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0412346566488585e-05,
+      "loss": 0.0086,
+      "step": 20471
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0402584537396316e-05,
+      "loss": 0.001,
+      "step": 20472
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.039282467282551e-05,
+      "loss": 0.0019,
+      "step": 20473
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.03830669729392e-05,
+      "loss": 0.0019,
+      "step": 20474
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0373311437900323e-05,
+      "loss": 0.0022,
+      "step": 20475
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0363558067871853e-05,
+      "loss": 0.0003,
+      "step": 20476
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0353806863016657e-05,
+      "loss": 0.0026,
+      "step": 20477
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0344057823497604e-05,
+      "loss": 0.0004,
+      "step": 20478
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.033431094947755e-05,
+      "loss": 0.0031,
+      "step": 20479
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0324566241119244e-05,
+      "loss": 0.0047,
+      "step": 20480
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0314823698585507e-05,
+      "loss": 0.0001,
+      "step": 20481
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0305083322038928e-05,
+      "loss": 0.0006,
+      "step": 20482
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.029534511164236e-05,
+      "loss": 0.0035,
+      "step": 20483
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.02856090675583e-05,
+      "loss": 0.0068,
+      "step": 20484
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0275875189949475e-05,
+      "loss": 0.0001,
+      "step": 20485
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0266143478978368e-05,
+      "loss": 0.0018,
+      "step": 20486
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.025641393480757e-05,
+      "loss": 0.0035,
+      "step": 20487
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0246686557599562e-05,
+      "loss": 0.0019,
+      "step": 20488
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0236961347516818e-05,
+      "loss": 0.0004,
+      "step": 20489
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0227238304721755e-05,
+      "loss": 0.0014,
+      "step": 20490
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0217517429376777e-05,
+      "loss": 0.0011,
+      "step": 20491
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0207798721644242e-05,
+      "loss": 0.0004,
+      "step": 20492
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.019808218168648e-05,
+      "loss": 0.0002,
+      "step": 20493
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0188367809665785e-05,
+      "loss": 0.0012,
+      "step": 20494
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.017865560574432e-05,
+      "loss": 0.0006,
+      "step": 20495
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0168945570084428e-05,
+      "loss": 0.0009,
+      "step": 20496
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.015923770284816e-05,
+      "loss": 0.0013,
+      "step": 20497
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0149532004197765e-05,
+      "loss": 0.0004,
+      "step": 20498
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0139828474295272e-05,
+      "loss": 0.0005,
+      "step": 20499
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0130127113302745e-05,
+      "loss": 0.0015,
+      "step": 20500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0120427921382293e-05,
+      "loss": 0.0002,
+      "step": 20501
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0110730898695813e-05,
+      "loss": 0.0002,
+      "step": 20502
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0101036045405365e-05,
+      "loss": 0.002,
+      "step": 20503
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0091343361672785e-05,
+      "loss": 0.0062,
+      "step": 20504
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.008165284765999e-05,
+      "loss": 0.0014,
+      "step": 20505
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.007196450352882e-05,
+      "loss": 0.0007,
+      "step": 20506
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.006227832944111e-05,
+      "loss": 0.0028,
+      "step": 20507
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0052594325558637e-05,
+      "loss": 0.0017,
+      "step": 20508
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0042912492043113e-05,
+      "loss": 0.0016,
+      "step": 20509
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.003323282905628e-05,
+      "loss": 0.0009,
+      "step": 20510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0023555336759774e-05,
+      "loss": 0.0008,
+      "step": 20511
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.001388001531528e-05,
+      "loss": 0.0031,
+      "step": 20512
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0004206864884292e-05,
+      "loss": 0.0001,
+      "step": 20513
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9994535885628494e-05,
+      "loss": 0.0016,
+      "step": 20514
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9984867077709288e-05,
+      "loss": 0.0031,
+      "step": 20515
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.997520044128828e-05,
+      "loss": 0.0002,
+      "step": 20516
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9965535976526842e-05,
+      "loss": 0.0001,
+      "step": 20517
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.995587368358641e-05,
+      "loss": 0.0005,
+      "step": 20518
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9946213562628355e-05,
+      "loss": 0.0041,
+      "step": 20519
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.993655561381403e-05,
+      "loss": 0.0025,
+      "step": 20520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9926899837304745e-05,
+      "loss": 0.0004,
+      "step": 20521
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9917246233261763e-05,
+      "loss": 0.0002,
+      "step": 20522
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.990759480184631e-05,
+      "loss": 0.0014,
+      "step": 20523
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.989794554321959e-05,
+      "loss": 0.0003,
+      "step": 20524
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.988829845754277e-05,
+      "loss": 0.0014,
+      "step": 20525
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9878653544976964e-05,
+      "loss": 0.0062,
+      "step": 20526
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.98690108056833e-05,
+      "loss": 0.0032,
+      "step": 20527
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9859370239822735e-05,
+      "loss": 0.002,
+      "step": 20528
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.984973184755638e-05,
+      "loss": 0.0003,
+      "step": 20529
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9840095629045162e-05,
+      "loss": 0.0019,
+      "step": 20530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.983046158445006e-05,
+      "loss": 0.0004,
+      "step": 20531
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.982082971393199e-05,
+      "loss": 0.0002,
+      "step": 20532
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9811200017651714e-05,
+      "loss": 0.0029,
+      "step": 20533
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.980157249577023e-05,
+      "loss": 0.0064,
+      "step": 20534
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.979194714844818e-05,
+      "loss": 0.0007,
+      "step": 20535
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9782323975846463e-05,
+      "loss": 0.0003,
+      "step": 20536
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.977270297812571e-05,
+      "loss": 0.0025,
+      "step": 20537
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9763084155446636e-05,
+      "loss": 0.0004,
+      "step": 20538
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9753467507969893e-05,
+      "loss": 0.0048,
+      "step": 20539
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.974385303585611e-05,
+      "loss": 0.0001,
+      "step": 20540
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9734240739265837e-05,
+      "loss": 0.0003,
+      "step": 20541
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9724630618359665e-05,
+      "loss": 0.0015,
+      "step": 20542
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9715022673298047e-05,
+      "loss": 0.0001,
+      "step": 20543
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9705416904241505e-05,
+      "loss": 0.0025,
+      "step": 20544
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.969581331135046e-05,
+      "loss": 0.0015,
+      "step": 20545
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9686211894785254e-05,
+      "loss": 0.0006,
+      "step": 20546
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9676612654706354e-05,
+      "loss": 0.0008,
+      "step": 20547
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.966701559127395e-05,
+      "loss": 0.0022,
+      "step": 20548
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.965742070464848e-05,
+      "loss": 0.0007,
+      "step": 20549
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9647827994990083e-05,
+      "loss": 0.0017,
+      "step": 20550
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9638237462459016e-05,
+      "loss": 0.0011,
+      "step": 20551
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9628649107215466e-05,
+      "loss": 0.0011,
+      "step": 20552
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.961906292941954e-05,
+      "loss": 0.0019,
+      "step": 20553
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9609478929231436e-05,
+      "loss": 0.0002,
+      "step": 20554
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.959989710681112e-05,
+      "loss": 0.0038,
+      "step": 20555
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9590317462318688e-05,
+      "loss": 0.0019,
+      "step": 20556
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9580739995914103e-05,
+      "loss": 0.0004,
+      "step": 20557
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9571164707757347e-05,
+      "loss": 0.0024,
+      "step": 20558
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.956159159800835e-05,
+      "loss": 0.0013,
+      "step": 20559
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.955202066682699e-05,
+      "loss": 0.0,
+      "step": 20560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.954245191437313e-05,
+      "loss": 0.0003,
+      "step": 20561
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9532885340806576e-05,
+      "loss": 0.0014,
+      "step": 20562
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.952332094628713e-05,
+      "loss": 0.001,
+      "step": 20563
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9513758730974466e-05,
+      "loss": 0.0001,
+      "step": 20564
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9504198695028406e-05,
+      "loss": 0.0022,
+      "step": 20565
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.949464083860847e-05,
+      "loss": 0.01,
+      "step": 20566
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.948508516187447e-05,
+      "loss": 0.0013,
+      "step": 20567
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9475531664985873e-05,
+      "loss": 0.0006,
+      "step": 20568
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9465980348102266e-05,
+      "loss": 0.0026,
+      "step": 20569
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.945643121138321e-05,
+      "loss": 0.0061,
+      "step": 20570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.944688425498816e-05,
+      "loss": 0.0001,
+      "step": 20571
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9437339479076568e-05,
+      "loss": 0.0002,
+      "step": 20572
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.942779688380788e-05,
+      "loss": 0.0002,
+      "step": 20573
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9418256469341444e-05,
+      "loss": 0.0001,
+      "step": 20574
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9408718235836617e-05,
+      "loss": 0.0001,
+      "step": 20575
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.939918218345271e-05,
+      "loss": 0.003,
+      "step": 20576
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9389648312348976e-05,
+      "loss": 0.0001,
+      "step": 20577
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9380116622684706e-05,
+      "loss": 0.0006,
+      "step": 20578
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9370587114618986e-05,
+      "loss": 0.0218,
+      "step": 20579
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.936105978831109e-05,
+      "loss": 0.0002,
+      "step": 20580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9351534643920043e-05,
+      "loss": 0.0006,
+      "step": 20581
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9342011681605013e-05,
+      "loss": 0.0021,
+      "step": 20582
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.933249090152506e-05,
+      "loss": 0.0024,
+      "step": 20583
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.932297230383909e-05,
+      "loss": 0.0005,
+      "step": 20584
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9313455888706207e-05,
+      "loss": 0.004,
+      "step": 20585
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9303941656285287e-05,
+      "loss": 0.0033,
+      "step": 20586
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9294429606735235e-05,
+      "loss": 0.0022,
+      "step": 20587
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9284919740214938e-05,
+      "loss": 0.0044,
+      "step": 20588
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.927541205688322e-05,
+      "loss": 0.003,
+      "step": 20589
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9265906556898892e-05,
+      "loss": 0.0018,
+      "step": 20590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9256403240420686e-05,
+      "loss": 0.0016,
+      "step": 20591
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.924690210760735e-05,
+      "loss": 0.0007,
+      "step": 20592
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9237403158617566e-05,
+      "loss": 0.0024,
+      "step": 20593
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9227906393609975e-05,
+      "loss": 0.0018,
+      "step": 20594
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9218411812743206e-05,
+      "loss": 0.0043,
+      "step": 20595
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9208919416175846e-05,
+      "loss": 0.0005,
+      "step": 20596
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.919942920406636e-05,
+      "loss": 0.0013,
+      "step": 20597
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.918994117657337e-05,
+      "loss": 0.0014,
+      "step": 20598
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.918045533385522e-05,
+      "loss": 0.0008,
+      "step": 20599
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.917097167607049e-05,
+      "loss": 0.0004,
+      "step": 20600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9161490203377434e-05,
+      "loss": 0.0012,
+      "step": 20601
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9152010915934473e-05,
+      "loss": 0.0009,
+      "step": 20602
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9142533813899917e-05,
+      "loss": 0.0075,
+      "step": 20603
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9133058897432075e-05,
+      "loss": 0.001,
+      "step": 20604
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9123586166689165e-05,
+      "loss": 0.002,
+      "step": 20605
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9114115621829412e-05,
+      "loss": 0.0018,
+      "step": 20606
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.910464726301101e-05,
+      "loss": 0.0001,
+      "step": 20607
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9095181090392057e-05,
+      "loss": 0.0025,
+      "step": 20608
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.908571710413069e-05,
+      "loss": 0.0002,
+      "step": 20609
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9076255304384968e-05,
+      "loss": 0.0014,
+      "step": 20610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.906679569131294e-05,
+      "loss": 0.0061,
+      "step": 20611
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.905733826507253e-05,
+      "loss": 0.0065,
+      "step": 20612
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9047883025821774e-05,
+      "loss": 0.0017,
+      "step": 20613
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9038429973718583e-05,
+      "loss": 0.0002,
+      "step": 20614
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9028979108920762e-05,
+      "loss": 0.0007,
+      "step": 20615
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9019530431586267e-05,
+      "loss": 0.0012,
+      "step": 20616
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9010083941872807e-05,
+      "loss": 0.0001,
+      "step": 20617
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.900063963993827e-05,
+      "loss": 0.0013,
+      "step": 20618
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8991197525940284e-05,
+      "loss": 0.0003,
+      "step": 20619
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8981757600036602e-05,
+      "loss": 0.0009,
+      "step": 20620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8972319862384884e-05,
+      "loss": 0.0016,
+      "step": 20621
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.896288431314277e-05,
+      "loss": 0.0001,
+      "step": 20622
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8953450952467814e-05,
+      "loss": 0.0002,
+      "step": 20623
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.894401978051761e-05,
+      "loss": 0.0075,
+      "step": 20624
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.893459079744965e-05,
+      "loss": 0.0052,
+      "step": 20625
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.892516400342144e-05,
+      "loss": 0.004,
+      "step": 20626
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8915739398590418e-05,
+      "loss": 0.0003,
+      "step": 20627
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8906316983113963e-05,
+      "loss": 0.0053,
+      "step": 20628
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8896896757149525e-05,
+      "loss": 0.0002,
+      "step": 20629
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8887478720854303e-05,
+      "loss": 0.0058,
+      "step": 20630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.887806287438577e-05,
+      "loss": 0.0024,
+      "step": 20631
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.886864921790101e-05,
+      "loss": 0.0004,
+      "step": 20632
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8859237751557404e-05,
+      "loss": 0.0003,
+      "step": 20633
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8849828475512042e-05,
+      "loss": 0.0001,
+      "step": 20634
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8840421389922077e-05,
+      "loss": 0.0006,
+      "step": 20635
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8831016494944723e-05,
+      "loss": 0.0001,
+      "step": 20636
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8821613790736968e-05,
+      "loss": 0.0011,
+      "step": 20637
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.881221327745587e-05,
+      "loss": 0.0073,
+      "step": 20638
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8802814955258432e-05,
+      "loss": 0.0004,
+      "step": 20639
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8793418824301654e-05,
+      "loss": 0.002,
+      "step": 20640
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8784024884742455e-05,
+      "loss": 0.0016,
+      "step": 20641
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.877463313673771e-05,
+      "loss": 0.0056,
+      "step": 20642
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.876524358044431e-05,
+      "loss": 0.0012,
+      "step": 20643
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8755856216019067e-05,
+      "loss": 0.0015,
+      "step": 20644
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.874647104361875e-05,
+      "loss": 0.0009,
+      "step": 20645
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8737088063400136e-05,
+      "loss": 0.0001,
+      "step": 20646
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8727707275519966e-05,
+      "loss": 0.0001,
+      "step": 20647
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8718328680134815e-05,
+      "loss": 0.0046,
+      "step": 20648
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.870895227740144e-05,
+      "loss": 0.0009,
+      "step": 20649
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8699578067476335e-05,
+      "loss": 0.001,
+      "step": 20650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8690206050516187e-05,
+      "loss": 0.0024,
+      "step": 20651
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8680836226677425e-05,
+      "loss": 0.0008,
+      "step": 20652
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8671468596116572e-05,
+      "loss": 0.0009,
+      "step": 20653
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8662103158990104e-05,
+      "loss": 0.0007,
+      "step": 20654
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8652739915454444e-05,
+      "loss": 0.0005,
+      "step": 20655
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8643378865665953e-05,
+      "loss": 0.0002,
+      "step": 20656
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8634020009780987e-05,
+      "loss": 0.0024,
+      "step": 20657
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8624663347955853e-05,
+      "loss": 0.003,
+      "step": 20658
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.861530888034683e-05,
+      "loss": 0.0075,
+      "step": 20659
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8605956607110158e-05,
+      "loss": 0.0018,
+      "step": 20660
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.859660652840204e-05,
+      "loss": 0.0009,
+      "step": 20661
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.858725864437866e-05,
+      "loss": 0.0018,
+      "step": 20662
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8577912955196056e-05,
+      "loss": 0.0001,
+      "step": 20663
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.856856946101045e-05,
+      "loss": 0.0011,
+      "step": 20664
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8559228161977792e-05,
+      "loss": 0.0004,
+      "step": 20665
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8549889058254098e-05,
+      "loss": 0.0018,
+      "step": 20666
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8540552149995463e-05,
+      "loss": 0.0067,
+      "step": 20667
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8531217437357683e-05,
+      "loss": 0.0024,
+      "step": 20668
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.852188492049681e-05,
+      "loss": 0.0007,
+      "step": 20669
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8512554599568606e-05,
+      "loss": 0.0052,
+      "step": 20670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8503226474728933e-05,
+      "loss": 0.0019,
+      "step": 20671
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8493900546133596e-05,
+      "loss": 0.0029,
+      "step": 20672
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8484576813938367e-05,
+      "loss": 0.0012,
+      "step": 20673
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8475255278298946e-05,
+      "loss": 0.0003,
+      "step": 20674
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8465935939371036e-05,
+      "loss": 0.0005,
+      "step": 20675
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8456618797310278e-05,
+      "loss": 0.0019,
+      "step": 20676
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8447303852272304e-05,
+      "loss": 0.0001,
+      "step": 20677
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8437991104412668e-05,
+      "loss": 0.0019,
+      "step": 20678
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8428680553886934e-05,
+      "loss": 0.0021,
+      "step": 20679
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.841937220085062e-05,
+      "loss": 0.0018,
+      "step": 20680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8410066045459088e-05,
+      "loss": 0.0034,
+      "step": 20681
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8400762087867933e-05,
+      "loss": 0.0005,
+      "step": 20682
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8391460328232378e-05,
+      "loss": 0.0014,
+      "step": 20683
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8382160766707948e-05,
+      "loss": 0.0001,
+      "step": 20684
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8372863403449823e-05,
+      "loss": 0.0007,
+      "step": 20685
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8363568238613356e-05,
+      "loss": 0.0035,
+      "step": 20686
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8354275272353798e-05,
+      "loss": 0.0028,
+      "step": 20687
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8344984504826315e-05,
+      "loss": 0.0011,
+      "step": 20688
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8335695936186124e-05,
+      "loss": 0.0022,
+      "step": 20689
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8326409566588347e-05,
+      "loss": 0.003,
+      "step": 20690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.831712539618806e-05,
+      "loss": 0.0002,
+      "step": 20691
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8307843425140373e-05,
+      "loss": 0.0013,
+      "step": 20692
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8298563653600295e-05,
+      "loss": 0.0003,
+      "step": 20693
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8289286081722772e-05,
+      "loss": 0.0007,
+      "step": 20694
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8280010709662806e-05,
+      "loss": 0.0001,
+      "step": 20695
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8270737537575297e-05,
+      "loss": 0.0015,
+      "step": 20696
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8261466565615136e-05,
+      "loss": 0.0013,
+      "step": 20697
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8252197793937195e-05,
+      "loss": 0.0002,
+      "step": 20698
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8242931222696172e-05,
+      "loss": 0.0005,
+      "step": 20699
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8233666852046974e-05,
+      "loss": 0.0001,
+      "step": 20700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8224404682144193e-05,
+      "loss": 0.0002,
+      "step": 20701
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8215144713142678e-05,
+      "loss": 0.0003,
+      "step": 20702
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8205886945196964e-05,
+      "loss": 0.0006,
+      "step": 20703
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8196631378461713e-05,
+      "loss": 0.0061,
+      "step": 20704
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.818737801309152e-05,
+      "loss": 0.0001,
+      "step": 20705
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8178126849240927e-05,
+      "loss": 0.0007,
+      "step": 20706
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.816887788706444e-05,
+      "loss": 0.0042,
+      "step": 20707
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8159631126716562e-05,
+      "loss": 0.0022,
+      "step": 20708
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8150386568351694e-05,
+      "loss": 0.0015,
+      "step": 20709
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8141144212124265e-05,
+      "loss": 0.0011,
+      "step": 20710
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8131904058188652e-05,
+      "loss": 0.0002,
+      "step": 20711
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8122666106699096e-05,
+      "loss": 0.0002,
+      "step": 20712
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.811343035781001e-05,
+      "loss": 0.0012,
+      "step": 20713
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8104196811675537e-05,
+      "loss": 0.0033,
+      "step": 20714
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.809496546845002e-05,
+      "loss": 0.0018,
+      "step": 20715
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8085736328287514e-05,
+      "loss": 0.0008,
+      "step": 20716
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8076509391342236e-05,
+      "loss": 0.0038,
+      "step": 20717
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8067284657768262e-05,
+      "loss": 0.0018,
+      "step": 20718
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8058062127719648e-05,
+      "loss": 0.0041,
+      "step": 20719
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8048841801350524e-05,
+      "loss": 0.0024,
+      "step": 20720
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.803962367881478e-05,
+      "loss": 0.0007,
+      "step": 20721
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8030407760266414e-05,
+      "loss": 0.0035,
+      "step": 20722
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.802119404585935e-05,
+      "loss": 0.0004,
+      "step": 20723
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8011982535747465e-05,
+      "loss": 0.0006,
+      "step": 20724
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.800277323008462e-05,
+      "loss": 0.0001,
+      "step": 20725
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.799356612902463e-05,
+      "loss": 0.0016,
+      "step": 20726
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7984361232721262e-05,
+      "loss": 0.0026,
+      "step": 20727
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7975158541328266e-05,
+      "loss": 0.0025,
+      "step": 20728
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7965958054999336e-05,
+      "loss": 0.0025,
+      "step": 20729
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.795675977388813e-05,
+      "loss": 0.0002,
+      "step": 20730
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.794756369814831e-05,
+      "loss": 0.0015,
+      "step": 20731
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.79383698279334e-05,
+      "loss": 0.0001,
+      "step": 20732
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.792917816339705e-05,
+      "loss": 0.0012,
+      "step": 20733
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.791998870469265e-05,
+      "loss": 0.0017,
+      "step": 20734
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7910801451973844e-05,
+      "loss": 0.0007,
+      "step": 20735
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.790161640539394e-05,
+      "loss": 0.0013,
+      "step": 20736
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7892433565106408e-05,
+      "loss": 0.0003,
+      "step": 20737
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7883252931264598e-05,
+      "loss": 0.0002,
+      "step": 20738
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7874074504021857e-05,
+      "loss": 0.0005,
+      "step": 20739
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7864898283531482e-05,
+      "loss": 0.0047,
+      "step": 20740
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7855724269946715e-05,
+      "loss": 0.0003,
+      "step": 20741
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.78465524634208e-05,
+      "loss": 0.0018,
+      "step": 20742
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7837382864106913e-05,
+      "loss": 0.0017,
+      "step": 20743
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7828215472158253e-05,
+      "loss": 0.0079,
+      "step": 20744
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.781905028772781e-05,
+      "loss": 0.0053,
+      "step": 20745
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7809887310968795e-05,
+      "loss": 0.0112,
+      "step": 20746
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.780072654203414e-05,
+      "loss": 0.0006,
+      "step": 20747
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7791567981076915e-05,
+      "loss": 0.0011,
+      "step": 20748
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7782411628250088e-05,
+      "loss": 0.0037,
+      "step": 20749
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7773257483706517e-05,
+      "loss": 0.0004,
+      "step": 20750
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.776410554759921e-05,
+      "loss": 0.0003,
+      "step": 20751
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7754955820080886e-05,
+      "loss": 0.0016,
+      "step": 20752
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.774580830130448e-05,
+      "loss": 0.0018,
+      "step": 20753
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7736662991422712e-05,
+      "loss": 0.0005,
+      "step": 20754
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.772751989058832e-05,
+      "loss": 0.0011,
+      "step": 20755
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7718378998954035e-05,
+      "loss": 0.0012,
+      "step": 20756
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.770924031667252e-05,
+      "loss": 0.0005,
+      "step": 20757
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7700103843896413e-05,
+      "loss": 0.0001,
+      "step": 20758
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7690969580778297e-05,
+      "loss": 0.0022,
+      "step": 20759
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7681837527470732e-05,
+      "loss": 0.0037,
+      "step": 20760
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7672707684126265e-05,
+      "loss": 0.0097,
+      "step": 20761
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7663580050897387e-05,
+      "loss": 0.0002,
+      "step": 20762
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7654454627936477e-05,
+      "loss": 0.0031,
+      "step": 20763
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.764533141539603e-05,
+      "loss": 0.0104,
+      "step": 20764
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.763621041342834e-05,
+      "loss": 0.0011,
+      "step": 20765
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7627091622185857e-05,
+      "loss": 0.0001,
+      "step": 20766
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7617975041820786e-05,
+      "loss": 0.0002,
+      "step": 20767
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7608860672485407e-05,
+      "loss": 0.0014,
+      "step": 20768
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.759974851433197e-05,
+      "loss": 0.0029,
+      "step": 20769
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7590638567512662e-05,
+      "loss": 0.0023,
+      "step": 20770
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7581530832179614e-05,
+      "loss": 0.0043,
+      "step": 20771
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.757242530848497e-05,
+      "loss": 0.0024,
+      "step": 20772
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7563321996580798e-05,
+      "loss": 0.0026,
+      "step": 20773
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7554220896619137e-05,
+      "loss": 0.0055,
+      "step": 20774
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7545122008751984e-05,
+      "loss": 0.0043,
+      "step": 20775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.753602533313133e-05,
+      "loss": 0.0011,
+      "step": 20776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7526930869909144e-05,
+      "loss": 0.0042,
+      "step": 20777
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7517838619237186e-05,
+      "loss": 0.0007,
+      "step": 20778
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7508748581267444e-05,
+      "loss": 0.0005,
+      "step": 20779
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.749966075615169e-05,
+      "loss": 0.0039,
+      "step": 20780
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.749057514404173e-05,
+      "loss": 0.0057,
+      "step": 20781
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.748149174508931e-05,
+      "loss": 0.0032,
+      "step": 20782
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.747241055944606e-05,
+      "loss": 0.0012,
+      "step": 20783
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.746333158726379e-05,
+      "loss": 0.0004,
+      "step": 20784
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.745425482869403e-05,
+      "loss": 0.0024,
+      "step": 20785
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7445180283888394e-05,
+      "loss": 0.0014,
+      "step": 20786
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.743610795299849e-05,
+      "loss": 0.0001,
+      "step": 20787
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.74270378361758e-05,
+      "loss": 0.0036,
+      "step": 20788
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7417969933571834e-05,
+      "loss": 0.0035,
+      "step": 20789
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7408904245338024e-05,
+      "loss": 0.0034,
+      "step": 20790
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.739984077162581e-05,
+      "loss": 0.0033,
+      "step": 20791
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7390779512586544e-05,
+      "loss": 0.0007,
+      "step": 20792
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7381720468371586e-05,
+      "loss": 0.0004,
+      "step": 20793
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.737266363913223e-05,
+      "loss": 0.0026,
+      "step": 20794
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.736360902501977e-05,
+      "loss": 0.0015,
+      "step": 20795
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7354556626185357e-05,
+      "loss": 0.0,
+      "step": 20796
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.734550644278028e-05,
+      "loss": 0.0022,
+      "step": 20797
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7336458474955596e-05,
+      "loss": 0.0024,
+      "step": 20798
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7327412722862538e-05,
+      "loss": 0.0021,
+      "step": 20799
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.731836918665209e-05,
+      "loss": 0.0004,
+      "step": 20800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7309327866475336e-05,
+      "loss": 0.0026,
+      "step": 20801
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7300288762483282e-05,
+      "loss": 0.0015,
+      "step": 20802
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7291251874826844e-05,
+      "loss": 0.0054,
+      "step": 20803
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7282217203657083e-05,
+      "loss": 0.0047,
+      "step": 20804
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.727318474912478e-05,
+      "loss": 0.0003,
+      "step": 20805
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7264154511380845e-05,
+      "loss": 0.0015,
+      "step": 20806
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7255126490576077e-05,
+      "loss": 0.0031,
+      "step": 20807
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.724610068686127e-05,
+      "loss": 0.0001,
+      "step": 20808
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.723707710038717e-05,
+      "loss": 0.0004,
+      "step": 20809
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.722805573130449e-05,
+      "loss": 0.0001,
+      "step": 20810
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7219036579763895e-05,
+      "loss": 0.0016,
+      "step": 20811
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7210019645916045e-05,
+      "loss": 0.001,
+      "step": 20812
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7201004929911555e-05,
+      "loss": 0.0006,
+      "step": 20813
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7191992431900903e-05,
+      "loss": 0.0003,
+      "step": 20814
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7182982152034718e-05,
+      "loss": 0.0004,
+      "step": 20815
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7173974090463384e-05,
+      "loss": 0.0007,
+      "step": 20816
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7164968247337475e-05,
+      "loss": 0.0013,
+      "step": 20817
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7155964622807305e-05,
+      "loss": 0.0004,
+      "step": 20818
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.714696321702329e-05,
+      "loss": 0.0069,
+      "step": 20819
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7137964030135758e-05,
+      "loss": 0.0007,
+      "step": 20820
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.712896706229504e-05,
+      "loss": 0.0009,
+      "step": 20821
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7119972313651366e-05,
+      "loss": 0.0008,
+      "step": 20822
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7110979784354995e-05,
+      "loss": 0.005,
+      "step": 20823
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7101989474556094e-05,
+      "loss": 0.0001,
+      "step": 20824
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7093001384404842e-05,
+      "loss": 0.0008,
+      "step": 20825
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7084015514051352e-05,
+      "loss": 0.0012,
+      "step": 20826
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.707503186364569e-05,
+      "loss": 0.0,
+      "step": 20827
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7066050433337945e-05,
+      "loss": 0.0002,
+      "step": 20828
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7057071223278024e-05,
+      "loss": 0.0001,
+      "step": 20829
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7048094233616017e-05,
+      "loss": 0.0003,
+      "step": 20830
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7039119464501738e-05,
+      "loss": 0.0023,
+      "step": 20831
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7030146916085185e-05,
+      "loss": 0.0025,
+      "step": 20832
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7021176588516205e-05,
+      "loss": 0.0024,
+      "step": 20833
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7012208481944543e-05,
+      "loss": 0.0032,
+      "step": 20834
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7003242596520085e-05,
+      "loss": 0.0015,
+      "step": 20835
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6994278932392487e-05,
+      "loss": 0.001,
+      "step": 20836
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6985317489711497e-05,
+      "loss": 0.0014,
+      "step": 20837
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6976358268626783e-05,
+      "loss": 0.0011,
+      "step": 20838
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6967401269288005e-05,
+      "loss": 0.0006,
+      "step": 20839
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.695844649184473e-05,
+      "loss": 0.0004,
+      "step": 20840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6949493936446533e-05,
+      "loss": 0.0007,
+      "step": 20841
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6940543603242936e-05,
+      "loss": 0.0009,
+      "step": 20842
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6931595492383426e-05,
+      "loss": 0.0105,
+      "step": 20843
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6922649604017457e-05,
+      "loss": 0.0007,
+      "step": 20844
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.691370593829446e-05,
+      "loss": 0.0001,
+      "step": 20845
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6904764495363793e-05,
+      "loss": 0.0007,
+      "step": 20846
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6895825275374758e-05,
+      "loss": 0.0052,
+      "step": 20847
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6886888278476735e-05,
+      "loss": 0.0015,
+      "step": 20848
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6877953504818902e-05,
+      "loss": 0.0022,
+      "step": 20849
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.686902095455058e-05,
+      "loss": 0.0031,
+      "step": 20850
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.686009062782089e-05,
+      "loss": 0.0001,
+      "step": 20851
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6851162524779e-05,
+      "loss": 0.001,
+      "step": 20852
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6842236645574026e-05,
+      "loss": 0.0002,
+      "step": 20853
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6833312990355062e-05,
+      "loss": 0.0001,
+      "step": 20854
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6824391559271128e-05,
+      "loss": 0.0031,
+      "step": 20855
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.681547235247125e-05,
+      "loss": 0.0031,
+      "step": 20856
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.680655537010439e-05,
+      "loss": 0.0003,
+      "step": 20857
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6797640612319487e-05,
+      "loss": 0.0043,
+      "step": 20858
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.678872807926543e-05,
+      "loss": 0.0009,
+      "step": 20859
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.677981777109102e-05,
+      "loss": 0.0004,
+      "step": 20860
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6770909687945188e-05,
+      "loss": 0.0002,
+      "step": 20861
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6762003829976585e-05,
+      "loss": 0.0013,
+      "step": 20862
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6753100197334052e-05,
+      "loss": 0.0002,
+      "step": 20863
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.674419879016631e-05,
+      "loss": 0.0003,
+      "step": 20864
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.67352996086219e-05,
+      "loss": 0.0004,
+      "step": 20865
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6726402652849623e-05,
+      "loss": 0.0031,
+      "step": 20866
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.671750792299793e-05,
+      "loss": 0.0034,
+      "step": 20867
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.67086154192155e-05,
+      "loss": 0.0006,
+      "step": 20868
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.669972514165076e-05,
+      "loss": 0.0008,
+      "step": 20869
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6690837090452247e-05,
+      "loss": 0.001,
+      "step": 20870
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.668195126576839e-05,
+      "loss": 0.0001,
+      "step": 20871
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6673067667747603e-05,
+      "loss": 0.0004,
+      "step": 20872
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.666418629653825e-05,
+      "loss": 0.0003,
+      "step": 20873
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6655307152288683e-05,
+      "loss": 0.0001,
+      "step": 20874
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6646430235147194e-05,
+      "loss": 0.0003,
+      "step": 20875
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.663755554526204e-05,
+      "loss": 0.0019,
+      "step": 20876
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.662868308278144e-05,
+      "loss": 0.0022,
+      "step": 20877
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6619812847853603e-05,
+      "loss": 0.0023,
+      "step": 20878
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6610944840626672e-05,
+      "loss": 0.0002,
+      "step": 20879
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6602079061248713e-05,
+      "loss": 0.0011,
+      "step": 20880
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.659321550986789e-05,
+      "loss": 0.0008,
+      "step": 20881
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6584354186632137e-05,
+      "loss": 0.0023,
+      "step": 20882
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6575495091689565e-05,
+      "loss": 0.0001,
+      "step": 20883
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.656663822518806e-05,
+      "loss": 0.0001,
+      "step": 20884
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6557783587275535e-05,
+      "loss": 0.0003,
+      "step": 20885
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.654893117809997e-05,
+      "loss": 0.0025,
+      "step": 20886
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6540080997809135e-05,
+      "loss": 0.0016,
+      "step": 20887
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6531233046550873e-05,
+      "loss": 0.0005,
+      "step": 20888
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6522387324472952e-05,
+      "loss": 0.0013,
+      "step": 20889
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.651354383172312e-05,
+      "loss": 0.0106,
+      "step": 20890
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6504702568449073e-05,
+      "loss": 0.0032,
+      "step": 20891
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6495863534798498e-05,
+      "loss": 0.0009,
+      "step": 20892
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6487026730919005e-05,
+      "loss": 0.0001,
+      "step": 20893
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.64781921569582e-05,
+      "loss": 0.008,
+      "step": 20894
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6469359813063605e-05,
+      "loss": 0.0002,
+      "step": 20895
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6460529699382757e-05,
+      "loss": 0.0001,
+      "step": 20896
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.645170181606319e-05,
+      "loss": 0.0003,
+      "step": 20897
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6442876163252217e-05,
+      "loss": 0.0005,
+      "step": 20898
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6434052741097387e-05,
+      "loss": 0.0024,
+      "step": 20899
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6425231549745933e-05,
+      "loss": 0.0005,
+      "step": 20900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.64164125893453e-05,
+      "loss": 0.0003,
+      "step": 20901
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6407595860042728e-05,
+      "loss": 0.0001,
+      "step": 20902
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6398781361985474e-05,
+      "loss": 0.0009,
+      "step": 20903
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6389969095320743e-05,
+      "loss": 0.0005,
+      "step": 20904
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6381159060195748e-05,
+      "loss": 0.0002,
+      "step": 20905
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.637235125675762e-05,
+      "loss": 0.0007,
+      "step": 20906
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.636354568515348e-05,
+      "loss": 0.0028,
+      "step": 20907
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6354742345530353e-05,
+      "loss": 0.0001,
+      "step": 20908
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6345941238035328e-05,
+      "loss": 0.0006,
+      "step": 20909
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6337142362815397e-05,
+      "loss": 0.0072,
+      "step": 20910
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6328345720017433e-05,
+      "loss": 0.0007,
+      "step": 20911
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.631955130978848e-05,
+      "loss": 0.0001,
+      "step": 20912
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6310759132275292e-05,
+      "loss": 0.0005,
+      "step": 20913
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6301969187624848e-05,
+      "loss": 0.0007,
+      "step": 20914
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6293181475983863e-05,
+      "loss": 0.0011,
+      "step": 20915
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6284395997499107e-05,
+      "loss": 0.0006,
+      "step": 20916
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6275612752317408e-05,
+      "loss": 0.0052,
+      "step": 20917
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6266831740585334e-05,
+      "loss": 0.0069,
+      "step": 20918
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6258052962449668e-05,
+      "loss": 0.0001,
+      "step": 20919
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6249276418056945e-05,
+      "loss": 0.0042,
+      "step": 20920
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6240502107553777e-05,
+      "loss": 0.0006,
+      "step": 20921
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6231730031086713e-05,
+      "loss": 0.0007,
+      "step": 20922
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.622296018880226e-05,
+      "loss": 0.0009,
+      "step": 20923
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6214192580846907e-05,
+      "loss": 0.0001,
+      "step": 20924
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6205427207367062e-05,
+      "loss": 0.0007,
+      "step": 20925
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6196664068509135e-05,
+      "loss": 0.0013,
+      "step": 20926
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6187903164419496e-05,
+      "loss": 0.0002,
+      "step": 20927
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6179144495244477e-05,
+      "loss": 0.0019,
+      "step": 20928
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6170388061130328e-05,
+      "loss": 0.0029,
+      "step": 20929
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6161633862223367e-05,
+      "loss": 0.003,
+      "step": 20930
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.615288189866969e-05,
+      "loss": 0.0016,
+      "step": 20931
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6144132170615598e-05,
+      "loss": 0.0018,
+      "step": 20932
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6135384678207142e-05,
+      "loss": 0.0008,
+      "step": 20933
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.612663942159044e-05,
+      "loss": 0.0001,
+      "step": 20934
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6117896400911566e-05,
+      "loss": 0.0025,
+      "step": 20935
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.610915561631655e-05,
+      "loss": 0.0001,
+      "step": 20936
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.610041706795138e-05,
+      "loss": 0.0006,
+      "step": 20937
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6091680755961983e-05,
+      "loss": 0.0001,
+      "step": 20938
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6082946680494304e-05,
+      "loss": 0.0015,
+      "step": 20939
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6074214841694188e-05,
+      "loss": 0.0085,
+      "step": 20940
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6065485239707478e-05,
+      "loss": 0.0024,
+      "step": 20941
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.605675787468e-05,
+      "loss": 0.0023,
+      "step": 20942
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.604803274675752e-05,
+      "loss": 0.0005,
+      "step": 20943
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.60393098560857e-05,
+      "loss": 0.0043,
+      "step": 20944
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.603058920281029e-05,
+      "loss": 0.0034,
+      "step": 20945
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6021870787076934e-05,
+      "loss": 0.0016,
+      "step": 20946
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6013154609031233e-05,
+      "loss": 0.0017,
+      "step": 20947
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6004440668818813e-05,
+      "loss": 0.0002,
+      "step": 20948
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.599572896658508e-05,
+      "loss": 0.0006,
+      "step": 20949
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5987019502475702e-05,
+      "loss": 0.0026,
+      "step": 20950
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5978312276636006e-05,
+      "loss": 0.0001,
+      "step": 20951
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5969607289211532e-05,
+      "loss": 0.0005,
+      "step": 20952
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5960904540347595e-05,
+      "loss": 0.0053,
+      "step": 20953
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5952204030189563e-05,
+      "loss": 0.0074,
+      "step": 20954
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5943505758882768e-05,
+      "loss": 0.0029,
+      "step": 20955
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.593480972657246e-05,
+      "loss": 0.0007,
+      "step": 20956
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.592611593340391e-05,
+      "loss": 0.0003,
+      "step": 20957
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5917424379522297e-05,
+      "loss": 0.0001,
+      "step": 20958
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5908735065072793e-05,
+      "loss": 0.0002,
+      "step": 20959
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5900047990200542e-05,
+      "loss": 0.0007,
+      "step": 20960
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5891363155050635e-05,
+      "loss": 0.0023,
+      "step": 20961
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.588268055976805e-05,
+      "loss": 0.0002,
+      "step": 20962
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5874000204497946e-05,
+      "loss": 0.0055,
+      "step": 20963
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5865322089385146e-05,
+      "loss": 0.0001,
+      "step": 20964
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5856646214574725e-05,
+      "loss": 0.0058,
+      "step": 20965
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5847972580211482e-05,
+      "loss": 0.0009,
+      "step": 20966
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.583930118644035e-05,
+      "loss": 0.0001,
+      "step": 20967
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5830632033406117e-05,
+      "loss": 0.0045,
+      "step": 20968
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5821965121253576e-05,
+      "loss": 0.0012,
+      "step": 20969
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5813300450127552e-05,
+      "loss": 0.0002,
+      "step": 20970
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.580463802017266e-05,
+      "loss": 0.0035,
+      "step": 20971
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.579597783153365e-05,
+      "loss": 0.0006,
+      "step": 20972
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5787319884355125e-05,
+      "loss": 0.0001,
+      "step": 20973
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5778664178781703e-05,
+      "loss": 0.0026,
+      "step": 20974
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5770010714957946e-05,
+      "loss": 0.0024,
+      "step": 20975
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5761359493028398e-05,
+      "loss": 0.0003,
+      "step": 20976
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5752710513137533e-05,
+      "loss": 0.0002,
+      "step": 20977
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.574406377542981e-05,
+      "loss": 0.0008,
+      "step": 20978
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5735419280049645e-05,
+      "loss": 0.0024,
+      "step": 20979
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5726777027141436e-05,
+      "loss": 0.0041,
+      "step": 20980
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.571813701684952e-05,
+      "loss": 0.0011,
+      "step": 20981
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5709499249318148e-05,
+      "loss": 0.0013,
+      "step": 20982
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5700863724691685e-05,
+      "loss": 0.0025,
+      "step": 20983
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5692230443114272e-05,
+      "loss": 0.0044,
+      "step": 20984
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5683599404730124e-05,
+      "loss": 0.0002,
+      "step": 20985
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5674970609683425e-05,
+      "loss": 0.0029,
+      "step": 20986
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5666344058118257e-05,
+      "loss": 0.0009,
+      "step": 20987
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5657719750178727e-05,
+      "loss": 0.0024,
+      "step": 20988
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5649097686008864e-05,
+      "loss": 0.0151,
+      "step": 20989
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5640477865752678e-05,
+      "loss": 0.0015,
+      "step": 20990
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5631860289554126e-05,
+      "loss": 0.002,
+      "step": 20991
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.562324495755715e-05,
+      "loss": 0.0013,
+      "step": 20992
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5614631869905635e-05,
+      "loss": 0.0031,
+      "step": 20993
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5606021026743466e-05,
+      "loss": 0.0011,
+      "step": 20994
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5597412428214373e-05,
+      "loss": 0.0013,
+      "step": 20995
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5588806074462262e-05,
+      "loss": 0.0018,
+      "step": 20996
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.558020196563075e-05,
+      "loss": 0.0004,
+      "step": 20997
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5571600101863625e-05,
+      "loss": 0.0031,
+      "step": 20998
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5563000483304548e-05,
+      "loss": 0.0001,
+      "step": 20999
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5554403110097098e-05,
+      "loss": 0.0042,
+      "step": 21000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5545807982384935e-05,
+      "loss": 0.0004,
+      "step": 21001
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.553721510031153e-05,
+      "loss": 0.0005,
+      "step": 21002
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.55286244640205e-05,
+      "loss": 0.0025,
+      "step": 21003
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5520036073655267e-05,
+      "loss": 0.0025,
+      "step": 21004
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5511449929359267e-05,
+      "loss": 0.0002,
+      "step": 21005
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5502866031275913e-05,
+      "loss": 0.0011,
+      "step": 21006
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5494284379548577e-05,
+      "loss": 0.0051,
+      "step": 21007
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5485704974320594e-05,
+      "loss": 0.0009,
+      "step": 21008
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.547712781573525e-05,
+      "loss": 0.0103,
+      "step": 21009
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.54685529039358e-05,
+      "loss": 0.0009,
+      "step": 21010
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5459980239065474e-05,
+      "loss": 0.0014,
+      "step": 21011
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.545140982126746e-05,
+      "loss": 0.0013,
+      "step": 21012
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5442841650684824e-05,
+      "loss": 0.0008,
+      "step": 21013
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5434275727460786e-05,
+      "loss": 0.0005,
+      "step": 21014
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5425712051738287e-05,
+      "loss": 0.001,
+      "step": 21015
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.541715062366049e-05,
+      "loss": 0.0009,
+      "step": 21016
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5408591443370284e-05,
+      "loss": 0.0029,
+      "step": 21017
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.540003451101067e-05,
+      "loss": 0.0009,
+      "step": 21018
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5391479826724562e-05,
+      "loss": 0.0031,
+      "step": 21019
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.538292739065482e-05,
+      "loss": 0.001,
+      "step": 21020
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5374377202944293e-05,
+      "loss": 0.0006,
+      "step": 21021
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.53658292637358e-05,
+      "loss": 0.0026,
+      "step": 21022
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5357283573172098e-05,
+      "loss": 0.0017,
+      "step": 21023
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5348740131395915e-05,
+      "loss": 0.0057,
+      "step": 21024
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.534019893854994e-05,
+      "loss": 0.0017,
+      "step": 21025
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5331659994776847e-05,
+      "loss": 0.0004,
+      "step": 21026
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5323123300219246e-05,
+      "loss": 0.0002,
+      "step": 21027
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5314588855019666e-05,
+      "loss": 0.0005,
+      "step": 21028
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5306056659320704e-05,
+      "loss": 0.0027,
+      "step": 21029
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5297526713264847e-05,
+      "loss": 0.0054,
+      "step": 21030
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.528899901699458e-05,
+      "loss": 0.0022,
+      "step": 21031
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5280473570652324e-05,
+      "loss": 0.0003,
+      "step": 21032
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5271950374380416e-05,
+      "loss": 0.0032,
+      "step": 21033
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5263429428321305e-05,
+      "loss": 0.0008,
+      "step": 21034
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5254910732617232e-05,
+      "loss": 0.0029,
+      "step": 21035
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5246394287410495e-05,
+      "loss": 0.0027,
+      "step": 21036
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5237880092843336e-05,
+      "loss": 0.0025,
+      "step": 21037
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5229368149057969e-05,
+      "loss": 0.0022,
+      "step": 21038
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5220858456196533e-05,
+      "loss": 0.0019,
+      "step": 21039
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5212351014401198e-05,
+      "loss": 0.0013,
+      "step": 21040
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5203845823814016e-05,
+      "loss": 0.0051,
+      "step": 21041
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5195342884577056e-05,
+      "loss": 0.0045,
+      "step": 21042
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5186842196832339e-05,
+      "loss": 0.0057,
+      "step": 21043
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5178343760721834e-05,
+      "loss": 0.0046,
+      "step": 21044
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5169847576387511e-05,
+      "loss": 0.003,
+      "step": 21045
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5161353643971174e-05,
+      "loss": 0.0047,
+      "step": 21046
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5152861963614827e-05,
+      "loss": 0.0002,
+      "step": 21047
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.514437253546017e-05,
+      "loss": 0.0031,
+      "step": 21048
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5135885359649108e-05,
+      "loss": 0.0006,
+      "step": 21049
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5127400436323295e-05,
+      "loss": 0.0013,
+      "step": 21050
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5118917765624467e-05,
+      "loss": 0.0021,
+      "step": 21051
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5110437347694377e-05,
+      "loss": 0.0067,
+      "step": 21052
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5101959182674544e-05,
+      "loss": 0.0005,
+      "step": 21053
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5093483270706674e-05,
+      "loss": 0.0002,
+      "step": 21054
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5085009611932287e-05,
+      "loss": 0.0006,
+      "step": 21055
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5076538206492883e-05,
+      "loss": 0.0019,
+      "step": 21056
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5068069054529985e-05,
+      "loss": 0.0058,
+      "step": 21057
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5059602156185047e-05,
+      "loss": 0.0008,
+      "step": 21058
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5051137511599454e-05,
+      "loss": 0.0046,
+      "step": 21059
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.504267512091461e-05,
+      "loss": 0.0001,
+      "step": 21060
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5034214984271837e-05,
+      "loss": 0.0009,
+      "step": 21061
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5025757101812452e-05,
+      "loss": 0.0003,
+      "step": 21062
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5017301473677713e-05,
+      "loss": 0.0021,
+      "step": 21063
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5008848100008803e-05,
+      "loss": 0.0011,
+      "step": 21064
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5000396980946978e-05,
+      "loss": 0.0053,
+      "step": 21065
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4991948116633307e-05,
+      "loss": 0.0011,
+      "step": 21066
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4983501507209012e-05,
+      "loss": 0.0002,
+      "step": 21067
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4975057152815061e-05,
+      "loss": 0.0047,
+      "step": 21068
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4966615053592545e-05,
+      "loss": 0.0002,
+      "step": 21069
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4958175209682449e-05,
+      "loss": 0.001,
+      "step": 21070
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4949737621225727e-05,
+      "loss": 0.0007,
+      "step": 21071
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4941302288363316e-05,
+      "loss": 0.0008,
+      "step": 21072
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4932869211236105e-05,
+      "loss": 0.0017,
+      "step": 21073
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4924438389984927e-05,
+      "loss": 0.0025,
+      "step": 21074
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4916009824750608e-05,
+      "loss": 0.0012,
+      "step": 21075
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4907583515673915e-05,
+      "loss": 0.0001,
+      "step": 21076
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4899159462895588e-05,
+      "loss": 0.0024,
+      "step": 21077
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.489073766655633e-05,
+      "loss": 0.0003,
+      "step": 21078
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4882318126796743e-05,
+      "loss": 0.0003,
+      "step": 21079
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4873900843757552e-05,
+      "loss": 0.0002,
+      "step": 21080
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.486548581757926e-05,
+      "loss": 0.002,
+      "step": 21081
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4857073048402402e-05,
+      "loss": 0.0038,
+      "step": 21082
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4848662536367583e-05,
+      "loss": 0.0004,
+      "step": 21083
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.484025428161516e-05,
+      "loss": 0.0008,
+      "step": 21084
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4831848284285686e-05,
+      "loss": 0.0001,
+      "step": 21085
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4823444544519463e-05,
+      "loss": 0.0059,
+      "step": 21086
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4815043062456883e-05,
+      "loss": 0.0018,
+      "step": 21087
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4806643838238247e-05,
+      "loss": 0.0002,
+      "step": 21088
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4798246872003878e-05,
+      "loss": 0.0,
+      "step": 21089
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4789852163893978e-05,
+      "loss": 0.0005,
+      "step": 21090
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4781459714048788e-05,
+      "loss": 0.0001,
+      "step": 21091
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4773069522608444e-05,
+      "loss": 0.0044,
+      "step": 21092
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4764681589713117e-05,
+      "loss": 0.005,
+      "step": 21093
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.475629591550288e-05,
+      "loss": 0.0027,
+      "step": 21094
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4747912500117787e-05,
+      "loss": 0.0014,
+      "step": 21095
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.473953134369789e-05,
+      "loss": 0.0003,
+      "step": 21096
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4731152446383083e-05,
+      "loss": 0.0007,
+      "step": 21097
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.472277580831343e-05,
+      "loss": 0.0004,
+      "step": 21098
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4714401429628725e-05,
+      "loss": 0.0001,
+      "step": 21099
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.470602931046892e-05,
+      "loss": 0.0002,
+      "step": 21100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4697659450973803e-05,
+      "loss": 0.0001,
+      "step": 21101
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4689291851283164e-05,
+      "loss": 0.0008,
+      "step": 21102
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4680926511536773e-05,
+      "loss": 0.0012,
+      "step": 21103
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4672563431874318e-05,
+      "loss": 0.0018,
+      "step": 21104
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4664202612435554e-05,
+      "loss": 0.0009,
+      "step": 21105
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4655844053360038e-05,
+      "loss": 0.0011,
+      "step": 21106
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4647487754787406e-05,
+      "loss": 0.0005,
+      "step": 21107
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4639133716857231e-05,
+      "loss": 0.0028,
+      "step": 21108
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4630781939709069e-05,
+      "loss": 0.0003,
+      "step": 21109
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4622432423482306e-05,
+      "loss": 0.0065,
+      "step": 21110
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4614085168316497e-05,
+      "loss": 0.0065,
+      "step": 21111
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4605740174351033e-05,
+      "loss": 0.0008,
+      "step": 21112
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4597397441725284e-05,
+      "loss": 0.0003,
+      "step": 21113
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4589056970578605e-05,
+      "loss": 0.0001,
+      "step": 21114
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.458071876105022e-05,
+      "loss": 0.0002,
+      "step": 21115
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4572382813279532e-05,
+      "loss": 0.0001,
+      "step": 21116
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4564049127405614e-05,
+      "loss": 0.0005,
+      "step": 21117
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4555717703567787e-05,
+      "loss": 0.0003,
+      "step": 21118
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4547388541905125e-05,
+      "loss": 0.0006,
+      "step": 21119
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4539061642556749e-05,
+      "loss": 0.0013,
+      "step": 21120
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4530737005661747e-05,
+      "loss": 0.0011,
+      "step": 21121
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4522414631359158e-05,
+      "loss": 0.0018,
+      "step": 21122
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4514094519787973e-05,
+      "loss": 0.0018,
+      "step": 21123
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.450577667108716e-05,
+      "loss": 0.0006,
+      "step": 21124
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4497461085395629e-05,
+      "loss": 0.0032,
+      "step": 21125
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4489147762852282e-05,
+      "loss": 0.0009,
+      "step": 21126
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4480836703595977e-05,
+      "loss": 0.0054,
+      "step": 21127
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4472527907765485e-05,
+      "loss": 0.0017,
+      "step": 21128
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4464221375499647e-05,
+      "loss": 0.0015,
+      "step": 21129
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4455917106937098e-05,
+      "loss": 0.0001,
+      "step": 21130
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4447615102216648e-05,
+      "loss": 0.0017,
+      "step": 21131
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4439315361476867e-05,
+      "loss": 0.0009,
+      "step": 21132
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.443101788485641e-05,
+      "loss": 0.0002,
+      "step": 21133
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4422722672493853e-05,
+      "loss": 0.0008,
+      "step": 21134
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4414429724527732e-05,
+      "loss": 0.0005,
+      "step": 21135
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.440613904109662e-05,
+      "loss": 0.0015,
+      "step": 21136
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4397850622338908e-05,
+      "loss": 0.0109,
+      "step": 21137
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.438956446839305e-05,
+      "loss": 0.0027,
+      "step": 21138
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4381280579397452e-05,
+      "loss": 0.0026,
+      "step": 21139
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4372998955490472e-05,
+      "loss": 0.003,
+      "step": 21140
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4364719596810431e-05,
+      "loss": 0.0011,
+      "step": 21141
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4356442503495586e-05,
+      "loss": 0.0069,
+      "step": 21142
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4348167675684209e-05,
+      "loss": 0.0064,
+      "step": 21143
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4339895113514488e-05,
+      "loss": 0.0002,
+      "step": 21144
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.43316248171246e-05,
+      "loss": 0.0009,
+      "step": 21145
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4323356786652661e-05,
+      "loss": 0.0005,
+      "step": 21146
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4315091022236802e-05,
+      "loss": 0.0002,
+      "step": 21147
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.430682752401499e-05,
+      "loss": 0.0118,
+      "step": 21148
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.429856629212537e-05,
+      "loss": 0.0004,
+      "step": 21149
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4290307326705775e-05,
+      "loss": 0.0002,
+      "step": 21150
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4282050627894265e-05,
+      "loss": 0.0001,
+      "step": 21151
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4273796195828679e-05,
+      "loss": 0.0002,
+      "step": 21152
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4265544030646909e-05,
+      "loss": 0.0054,
+      "step": 21153
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4257294132486758e-05,
+      "loss": 0.0039,
+      "step": 21154
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4249046501486017e-05,
+      "loss": 0.0024,
+      "step": 21155
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4240801137782459e-05,
+      "loss": 0.0038,
+      "step": 21156
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4232558041513791e-05,
+      "loss": 0.0051,
+      "step": 21157
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.422431721281767e-05,
+      "loss": 0.0003,
+      "step": 21158
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4216078651831752e-05,
+      "loss": 0.0036,
+      "step": 21159
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4207842358693677e-05,
+      "loss": 0.0001,
+      "step": 21160
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4199608333540886e-05,
+      "loss": 0.0039,
+      "step": 21161
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4191376576511033e-05,
+      "loss": 0.0005,
+      "step": 21162
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4183147087741492e-05,
+      "loss": 0.0004,
+      "step": 21163
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4174919867369789e-05,
+      "loss": 0.0036,
+      "step": 21164
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.416669491553336e-05,
+      "loss": 0.0042,
+      "step": 21165
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.415847223236945e-05,
+      "loss": 0.0003,
+      "step": 21166
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4150251818015545e-05,
+      "loss": 0.0003,
+      "step": 21167
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4142033672608805e-05,
+      "loss": 0.0015,
+      "step": 21168
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4133817796286617e-05,
+      "loss": 0.001,
+      "step": 21169
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4125604189186106e-05,
+      "loss": 0.0018,
+      "step": 21170
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.411739285144448e-05,
+      "loss": 0.0094,
+      "step": 21171
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4109183783198897e-05,
+      "loss": 0.0021,
+      "step": 21172
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.410097698458646e-05,
+      "loss": 0.001,
+      "step": 21173
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4092772455744227e-05,
+      "loss": 0.0003,
+      "step": 21174
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4084570196809242e-05,
+      "loss": 0.0044,
+      "step": 21175
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4076370207918492e-05,
+      "loss": 0.0057,
+      "step": 21176
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4068172489208934e-05,
+      "loss": 0.0003,
+      "step": 21177
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4059977040817494e-05,
+      "loss": 0.0012,
+      "step": 21178
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4051783862881027e-05,
+      "loss": 0.0056,
+      "step": 21179
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4043592955536426e-05,
+      "loss": 0.002,
+      "step": 21180
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4035404318920395e-05,
+      "loss": 0.0009,
+      "step": 21181
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4027217953169828e-05,
+      "loss": 0.0001,
+      "step": 21182
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4019033858421347e-05,
+      "loss": 0.0006,
+      "step": 21183
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4010852034811692e-05,
+      "loss": 0.0002,
+      "step": 21184
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4002672482477506e-05,
+      "loss": 0.0004,
+      "step": 21185
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.399449520155541e-05,
+      "loss": 0.0,
+      "step": 21186
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3986320192181965e-05,
+      "loss": 0.0028,
+      "step": 21187
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.397814745449371e-05,
+      "loss": 0.0043,
+      "step": 21188
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3969976988627169e-05,
+      "loss": 0.0064,
+      "step": 21189
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3961808794718765e-05,
+      "loss": 0.007,
+      "step": 21190
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3953642872904958e-05,
+      "loss": 0.0016,
+      "step": 21191
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.394547922332212e-05,
+      "loss": 0.0003,
+      "step": 21192
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3937317846106626e-05,
+      "loss": 0.0002,
+      "step": 21193
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3929158741394702e-05,
+      "loss": 0.0006,
+      "step": 21194
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.392100190932272e-05,
+      "loss": 0.0005,
+      "step": 21195
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3912847350026857e-05,
+      "loss": 0.0,
+      "step": 21196
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3904695063643334e-05,
+      "loss": 0.0037,
+      "step": 21197
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3896545050308327e-05,
+      "loss": 0.0007,
+      "step": 21198
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3888397310157879e-05,
+      "loss": 0.0009,
+      "step": 21199
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3880251843328177e-05,
+      "loss": 0.0003,
+      "step": 21200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3872108649955166e-05,
+      "loss": 0.0021,
+      "step": 21201
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3863967730174951e-05,
+      "loss": 0.0021,
+      "step": 21202
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.385582908412341e-05,
+      "loss": 0.0015,
+      "step": 21203
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3847692711936531e-05,
+      "loss": 0.0045,
+      "step": 21204
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3839558613750174e-05,
+      "loss": 0.0043,
+      "step": 21205
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3831426789700228e-05,
+      "loss": 0.001,
+      "step": 21206
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.382329723992247e-05,
+      "loss": 0.0005,
+      "step": 21207
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3815169964552709e-05,
+      "loss": 0.001,
+      "step": 21208
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3807044963726683e-05,
+      "loss": 0.0001,
+      "step": 21209
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3798922237580085e-05,
+      "loss": 0.0004,
+      "step": 21210
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3790801786248607e-05,
+      "loss": 0.0006,
+      "step": 21211
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3782683609867806e-05,
+      "loss": 0.0024,
+      "step": 21212
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3774567708573359e-05,
+      "loss": 0.0024,
+      "step": 21213
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3766454082500721e-05,
+      "loss": 0.0007,
+      "step": 21214
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.375834273178552e-05,
+      "loss": 0.0013,
+      "step": 21215
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3750233656563148e-05,
+      "loss": 0.0015,
+      "step": 21216
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3742126856969026e-05,
+      "loss": 0.003,
+      "step": 21217
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.373402233313865e-05,
+      "loss": 0.0007,
+      "step": 21218
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3725920085207259e-05,
+      "loss": 0.0037,
+      "step": 21219
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3717820113310312e-05,
+      "loss": 0.0027,
+      "step": 21220
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.370972241758297e-05,
+      "loss": 0.0007,
+      "step": 21221
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3701626998160553e-05,
+      "loss": 0.0024,
+      "step": 21222
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3693533855178223e-05,
+      "loss": 0.0001,
+      "step": 21223
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.368544298877119e-05,
+      "loss": 0.0021,
+      "step": 21224
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3677354399074558e-05,
+      "loss": 0.0005,
+      "step": 21225
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.366926808622344e-05,
+      "loss": 0.0023,
+      "step": 21226
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3661184050352892e-05,
+      "loss": 0.0005,
+      "step": 21227
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3653102291597923e-05,
+      "loss": 0.0025,
+      "step": 21228
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3645022810093558e-05,
+      "loss": 0.0003,
+      "step": 21229
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3636945605974641e-05,
+      "loss": 0.0014,
+      "step": 21230
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3628870679376181e-05,
+      "loss": 0.0003,
+      "step": 21231
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.362079803043295e-05,
+      "loss": 0.0021,
+      "step": 21232
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3612727659279876e-05,
+      "loss": 0.0016,
+      "step": 21233
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3604659566051685e-05,
+      "loss": 0.0004,
+      "step": 21234
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3596593750883134e-05,
+      "loss": 0.0197,
+      "step": 21235
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.358853021390895e-05,
+      "loss": 0.0007,
+      "step": 21236
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.358046895526379e-05,
+      "loss": 0.0007,
+      "step": 21237
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3572409975082332e-05,
+      "loss": 0.0007,
+      "step": 21238
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3564353273499151e-05,
+      "loss": 0.0006,
+      "step": 21239
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3556298850648789e-05,
+      "loss": 0.0026,
+      "step": 21240
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3548246706665806e-05,
+      "loss": 0.0002,
+      "step": 21241
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3540196841684675e-05,
+      "loss": 0.0009,
+      "step": 21242
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3532149255839842e-05,
+      "loss": 0.0001,
+      "step": 21243
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3524103949265746e-05,
+      "loss": 0.0001,
+      "step": 21244
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3516060922096683e-05,
+      "loss": 0.0019,
+      "step": 21245
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3508020174467077e-05,
+      "loss": 0.0004,
+      "step": 21246
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3499981706511138e-05,
+      "loss": 0.0011,
+      "step": 21247
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.349194551836319e-05,
+      "loss": 0.001,
+      "step": 21248
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3483911610157444e-05,
+      "loss": 0.0002,
+      "step": 21249
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3475879982028025e-05,
+      "loss": 0.0002,
+      "step": 21250
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3467850634109161e-05,
+      "loss": 0.0069,
+      "step": 21251
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3459823566534877e-05,
+      "loss": 0.0002,
+      "step": 21252
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3451798779439315e-05,
+      "loss": 0.0022,
+      "step": 21253
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3443776272956453e-05,
+      "loss": 0.0001,
+      "step": 21254
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3435756047220282e-05,
+      "loss": 0.0027,
+      "step": 21255
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3427738102364776e-05,
+      "loss": 0.0002,
+      "step": 21256
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3419722438523834e-05,
+      "loss": 0.0017,
+      "step": 21257
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3411709055831344e-05,
+      "loss": 0.0046,
+      "step": 21258
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.340369795442115e-05,
+      "loss": 0.0021,
+      "step": 21259
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3395689134427045e-05,
+      "loss": 0.0,
+      "step": 21260
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3387682595982774e-05,
+      "loss": 0.0013,
+      "step": 21261
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3379678339222111e-05,
+      "loss": 0.0029,
+      "step": 21262
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3371676364278666e-05,
+      "loss": 0.0044,
+      "step": 21263
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3363676671286184e-05,
+      "loss": 0.0062,
+      "step": 21264
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3355679260378154e-05,
+      "loss": 0.0004,
+      "step": 21265
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3347684131688274e-05,
+      "loss": 0.0005,
+      "step": 21266
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3339691285349985e-05,
+      "loss": 0.0018,
+      "step": 21267
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.333170072149683e-05,
+      "loss": 0.0023,
+      "step": 21268
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3323712440262236e-05,
+      "loss": 0.0001,
+      "step": 21269
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3315726441779629e-05,
+      "loss": 0.001,
+      "step": 21270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3307742726182452e-05,
+      "loss": 0.0005,
+      "step": 21271
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.329976129360395e-05,
+      "loss": 0.0033,
+      "step": 21272
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3291782144177499e-05,
+      "loss": 0.0002,
+      "step": 21273
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3283805278036326e-05,
+      "loss": 0.0015,
+      "step": 21274
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3275830695313688e-05,
+      "loss": 0.0034,
+      "step": 21275
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.326785839614275e-05,
+      "loss": 0.0071,
+      "step": 21276
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3259888380656702e-05,
+      "loss": 0.0007,
+      "step": 21277
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.325192064898859e-05,
+      "loss": 0.0001,
+      "step": 21278
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3243955201271555e-05,
+      "loss": 0.0017,
+      "step": 21279
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3235992037638642e-05,
+      "loss": 0.0001,
+      "step": 21280
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3228031158222762e-05,
+      "loss": 0.0016,
+      "step": 21281
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3220072563157008e-05,
+      "loss": 0.0035,
+      "step": 21282
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3212116252574156e-05,
+      "loss": 0.0016,
+      "step": 21283
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3204162226607235e-05,
+      "loss": 0.0012,
+      "step": 21284
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3196210485389003e-05,
+      "loss": 0.0032,
+      "step": 21285
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3188261029052289e-05,
+      "loss": 0.0011,
+      "step": 21286
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3180313857729852e-05,
+      "loss": 0.0036,
+      "step": 21287
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3172368971554453e-05,
+      "loss": 0.0005,
+      "step": 21288
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.316442637065877e-05,
+      "loss": 0.0012,
+      "step": 21289
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3156486055175464e-05,
+      "loss": 0.0004,
+      "step": 21290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3148548025237143e-05,
+      "loss": 0.0001,
+      "step": 21291
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3140612280976404e-05,
+      "loss": 0.0034,
+      "step": 21292
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3132678822525772e-05,
+      "loss": 0.0003,
+      "step": 21293
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3124747650017758e-05,
+      "loss": 0.002,
+      "step": 21294
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3116818763584874e-05,
+      "loss": 0.0007,
+      "step": 21295
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.310889216335943e-05,
+      "loss": 0.0005,
+      "step": 21296
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3100967849473954e-05,
+      "loss": 0.0086,
+      "step": 21297
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3093045822060655e-05,
+      "loss": 0.0002,
+      "step": 21298
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3085126081251995e-05,
+      "loss": 0.0023,
+      "step": 21299
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3077208627180136e-05,
+      "loss": 0.0008,
+      "step": 21300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3069293459977337e-05,
+      "loss": 0.0005,
+      "step": 21301
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.306138057977586e-05,
+      "loss": 0.0022,
+      "step": 21302
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3053469986707765e-05,
+      "loss": 0.0013,
+      "step": 21303
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3045561680905281e-05,
+      "loss": 0.0007,
+      "step": 21304
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3037655662500402e-05,
+      "loss": 0.0065,
+      "step": 21305
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.302975193162522e-05,
+      "loss": 0.0038,
+      "step": 21306
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3021850488411716e-05,
+      "loss": 0.0033,
+      "step": 21307
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3013951332991884e-05,
+      "loss": 0.0005,
+      "step": 21308
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.300605446549765e-05,
+      "loss": 0.0016,
+      "step": 21309
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2998159886060893e-05,
+      "loss": 0.0001,
+      "step": 21310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2990267594813474e-05,
+      "loss": 0.0026,
+      "step": 21311
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2982377591887204e-05,
+      "loss": 0.0013,
+      "step": 21312
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2974489877413912e-05,
+      "loss": 0.0016,
+      "step": 21313
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2966604451525225e-05,
+      "loss": 0.0033,
+      "step": 21314
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.295872131435297e-05,
+      "loss": 0.0033,
+      "step": 21315
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2950840466028694e-05,
+      "loss": 0.0015,
+      "step": 21316
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2942961906684141e-05,
+      "loss": 0.0,
+      "step": 21317
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2935085636450804e-05,
+      "loss": 0.002,
+      "step": 21318
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2927211655460279e-05,
+      "loss": 0.0024,
+      "step": 21319
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2919339963844061e-05,
+      "loss": 0.0032,
+      "step": 21320
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.291147056173361e-05,
+      "loss": 0.0012,
+      "step": 21321
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2903603449260374e-05,
+      "loss": 0.0006,
+      "step": 21322
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2895738626555763e-05,
+      "loss": 0.003,
+      "step": 21323
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2887876093751104e-05,
+      "loss": 0.0015,
+      "step": 21324
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2880015850977744e-05,
+      "loss": 0.0067,
+      "step": 21325
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2872157898366941e-05,
+      "loss": 0.0007,
+      "step": 21326
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2864302236049961e-05,
+      "loss": 0.0002,
+      "step": 21327
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2856448864158014e-05,
+      "loss": 0.0001,
+      "step": 21328
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2848597782822196e-05,
+      "loss": 0.0007,
+      "step": 21329
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2840748992173732e-05,
+      "loss": 0.0006,
+      "step": 21330
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2832902492343655e-05,
+      "loss": 0.0003,
+      "step": 21331
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2825058283462991e-05,
+      "loss": 0.0002,
+      "step": 21332
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2817216365662853e-05,
+      "loss": 0.0069,
+      "step": 21333
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2809376739074084e-05,
+      "loss": 0.0004,
+      "step": 21334
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2801539403827748e-05,
+      "loss": 0.0003,
+      "step": 21335
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2793704360054658e-05,
+      "loss": 0.002,
+      "step": 21336
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2785871607885707e-05,
+      "loss": 0.0002,
+      "step": 21337
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2778041147451706e-05,
+      "loss": 0.0032,
+      "step": 21338
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2770212978883437e-05,
+      "loss": 0.0036,
+      "step": 21339
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.276238710231166e-05,
+      "loss": 0.0039,
+      "step": 21340
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2754563517867072e-05,
+      "loss": 0.001,
+      "step": 21341
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2746742225680334e-05,
+      "loss": 0.0055,
+      "step": 21342
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2738923225882092e-05,
+      "loss": 0.0056,
+      "step": 21343
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2731106518602924e-05,
+      "loss": 0.0004,
+      "step": 21344
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2723292103973393e-05,
+      "loss": 0.0019,
+      "step": 21345
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2715479982124045e-05,
+      "loss": 0.0011,
+      "step": 21346
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2707670153185256e-05,
+      "loss": 0.0013,
+      "step": 21347
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.269986261728761e-05,
+      "loss": 0.0006,
+      "step": 21348
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2692057374561365e-05,
+      "loss": 0.0001,
+      "step": 21349
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2684254425137002e-05,
+      "loss": 0.0043,
+      "step": 21350
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.267645376914475e-05,
+      "loss": 0.0002,
+      "step": 21351
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2668655406714957e-05,
+      "loss": 0.0018,
+      "step": 21352
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2660859337977847e-05,
+      "loss": 0.0002,
+      "step": 21353
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2653065563063636e-05,
+      "loss": 0.0001,
+      "step": 21354
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.264527408210247e-05,
+      "loss": 0.0037,
+      "step": 21355
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2637484895224526e-05,
+      "loss": 0.0006,
+      "step": 21356
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2629698002559869e-05,
+      "loss": 0.0018,
+      "step": 21357
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2621913404238543e-05,
+      "loss": 0.0015,
+      "step": 21358
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.261413110039063e-05,
+      "loss": 0.0013,
+      "step": 21359
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2606351091146005e-05,
+      "loss": 0.0003,
+      "step": 21360
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2598573376634685e-05,
+      "loss": 0.0027,
+      "step": 21361
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2590797956986548e-05,
+      "loss": 0.0001,
+      "step": 21362
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2583024832331473e-05,
+      "loss": 0.0006,
+      "step": 21363
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2575254002799307e-05,
+      "loss": 0.0031,
+      "step": 21364
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2567485468519744e-05,
+      "loss": 0.0051,
+      "step": 21365
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2559719229622651e-05,
+      "loss": 0.0001,
+      "step": 21366
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2551955286237637e-05,
+      "loss": 0.0007,
+      "step": 21367
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2544193638494466e-05,
+      "loss": 0.004,
+      "step": 21368
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2536434286522702e-05,
+      "loss": 0.0026,
+      "step": 21369
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2528677230451972e-05,
+      "loss": 0.0017,
+      "step": 21370
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2520922470411827e-05,
+      "loss": 0.0008,
+      "step": 21371
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2513170006531792e-05,
+      "loss": 0.0002,
+      "step": 21372
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2505419838941332e-05,
+      "loss": 0.0054,
+      "step": 21373
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2497671967769896e-05,
+      "loss": 0.0,
+      "step": 21374
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.248992639314691e-05,
+      "loss": 0.0002,
+      "step": 21375
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2482183115201704e-05,
+      "loss": 0.0013,
+      "step": 21376
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2474442134063628e-05,
+      "loss": 0.0039,
+      "step": 21377
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2466703449861975e-05,
+      "loss": 0.0005,
+      "step": 21378
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.245896706272601e-05,
+      "loss": 0.0009,
+      "step": 21379
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2451232972784864e-05,
+      "loss": 0.0003,
+      "step": 21380
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2443501180167831e-05,
+      "loss": 0.0056,
+      "step": 21381
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2435771685003943e-05,
+      "loss": 0.0002,
+      "step": 21382
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2428044487422345e-05,
+      "loss": 0.0012,
+      "step": 21383
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2420319587552102e-05,
+      "loss": 0.0035,
+      "step": 21384
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2412596985522178e-05,
+      "loss": 0.0009,
+      "step": 21385
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2404876681461667e-05,
+      "loss": 0.0098,
+      "step": 21386
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2397158675499403e-05,
+      "loss": 0.0006,
+      "step": 21387
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2389442967764346e-05,
+      "loss": 0.0002,
+      "step": 21388
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2381729558385345e-05,
+      "loss": 0.0027,
+      "step": 21389
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.237401844749123e-05,
+      "loss": 0.0011,
+      "step": 21390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2366309635210797e-05,
+      "loss": 0.0012,
+      "step": 21391
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2358603121672794e-05,
+      "loss": 0.0023,
+      "step": 21392
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2350898907005934e-05,
+      "loss": 0.0045,
+      "step": 21393
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.23431969913389e-05,
+      "loss": 0.002,
+      "step": 21394
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2335497374800318e-05,
+      "loss": 0.0001,
+      "step": 21395
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2327800057518788e-05,
+      "loss": 0.002,
+      "step": 21396
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2320105039622891e-05,
+      "loss": 0.0056,
+      "step": 21397
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2312412321241088e-05,
+      "loss": 0.0052,
+      "step": 21398
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2304721902501962e-05,
+      "loss": 0.0004,
+      "step": 21399
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2297033783533828e-05,
+      "loss": 0.0001,
+      "step": 21400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2289347964465229e-05,
+      "loss": 0.0003,
+      "step": 21401
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2281664445424433e-05,
+      "loss": 0.0001,
+      "step": 21402
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2273983226539802e-05,
+      "loss": 0.0005,
+      "step": 21403
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2266304307939633e-05,
+      "loss": 0.0005,
+      "step": 21404
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.225862768975216e-05,
+      "loss": 0.0002,
+      "step": 21405
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2250953372105626e-05,
+      "loss": 0.0002,
+      "step": 21406
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2243281355128182e-05,
+      "loss": 0.0002,
+      "step": 21407
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2235611638947972e-05,
+      "loss": 0.0005,
+      "step": 21408
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2227944223693098e-05,
+      "loss": 0.002,
+      "step": 21409
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2220279109491654e-05,
+      "loss": 0.0001,
+      "step": 21410
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2212616296471572e-05,
+      "loss": 0.0007,
+      "step": 21411
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2204955784760934e-05,
+      "loss": 0.0003,
+      "step": 21412
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2197297574487585e-05,
+      "loss": 0.0008,
+      "step": 21413
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2189641665779526e-05,
+      "loss": 0.0032,
+      "step": 21414
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2181988058764603e-05,
+      "loss": 0.0011,
+      "step": 21415
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.217433675357058e-05,
+      "loss": 0.0012,
+      "step": 21416
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2166687750325355e-05,
+      "loss": 0.0044,
+      "step": 21417
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.215904104915656e-05,
+      "loss": 0.0102,
+      "step": 21418
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.215139665019204e-05,
+      "loss": 0.0005,
+      "step": 21419
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2143754553559365e-05,
+      "loss": 0.0004,
+      "step": 21420
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2136114759386212e-05,
+      "loss": 0.0008,
+      "step": 21421
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2128477267800196e-05,
+      "loss": 0.0003,
+      "step": 21422
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2120842078928833e-05,
+      "loss": 0.0024,
+      "step": 21423
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2113209192899686e-05,
+      "loss": 0.0018,
+      "step": 21424
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2105578609840238e-05,
+      "loss": 0.001,
+      "step": 21425
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2097950329877903e-05,
+      "loss": 0.0004,
+      "step": 21426
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2090324353140113e-05,
+      "loss": 0.0024,
+      "step": 21427
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2082700679754264e-05,
+      "loss": 0.0019,
+      "step": 21428
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2075079309847591e-05,
+      "loss": 0.0002,
+      "step": 21429
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2067460243547488e-05,
+      "loss": 0.0001,
+      "step": 21430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2059843480981124e-05,
+      "loss": 0.0077,
+      "step": 21431
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2052229022275795e-05,
+      "loss": 0.0048,
+      "step": 21432
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2044616867558631e-05,
+      "loss": 0.0017,
+      "step": 21433
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2037007016956752e-05,
+      "loss": 0.0019,
+      "step": 21434
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.20293994705973e-05,
+      "loss": 0.0002,
+      "step": 21435
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2021794228607295e-05,
+      "loss": 0.0002,
+      "step": 21436
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2014191291113784e-05,
+      "loss": 0.0012,
+      "step": 21437
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2006590658243764e-05,
+      "loss": 0.0011,
+      "step": 21438
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1998992330124135e-05,
+      "loss": 0.0046,
+      "step": 21439
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1991396306881845e-05,
+      "loss": 0.0022,
+      "step": 21440
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1983802588643743e-05,
+      "loss": 0.0029,
+      "step": 21441
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1976211175536659e-05,
+      "loss": 0.0018,
+      "step": 21442
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1968622067687412e-05,
+      "loss": 0.0008,
+      "step": 21443
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1961035265222679e-05,
+      "loss": 0.0033,
+      "step": 21444
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1953450768269263e-05,
+      "loss": 0.0031,
+      "step": 21445
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1945868576953776e-05,
+      "loss": 0.0007,
+      "step": 21446
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1938288691402903e-05,
+      "loss": 0.003,
+      "step": 21447
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1930711111743225e-05,
+      "loss": 0.0004,
+      "step": 21448
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1923135838101256e-05,
+      "loss": 0.0029,
+      "step": 21449
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1915562870603612e-05,
+      "loss": 0.0002,
+      "step": 21450
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.190799220937666e-05,
+      "loss": 0.0031,
+      "step": 21451
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1900423854546948e-05,
+      "loss": 0.0008,
+      "step": 21452
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1892857806240825e-05,
+      "loss": 0.0001,
+      "step": 21453
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1885294064584671e-05,
+      "loss": 0.0008,
+      "step": 21454
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1877732629704822e-05,
+      "loss": 0.0014,
+      "step": 21455
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1870173501727559e-05,
+      "loss": 0.001,
+      "step": 21456
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.186261668077913e-05,
+      "loss": 0.0057,
+      "step": 21457
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1855062166985768e-05,
+      "loss": 0.0003,
+      "step": 21458
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1847509960473623e-05,
+      "loss": 0.0004,
+      "step": 21459
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1839960061368842e-05,
+      "loss": 0.0002,
+      "step": 21460
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.183241246979756e-05,
+      "loss": 0.0007,
+      "step": 21461
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.182486718588574e-05,
+      "loss": 0.0008,
+      "step": 21462
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1817324209759515e-05,
+      "loss": 0.0033,
+      "step": 21463
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.180978354154477e-05,
+      "loss": 0.0017,
+      "step": 21464
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1802245181367537e-05,
+      "loss": 0.0041,
+      "step": 21465
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1794709129353646e-05,
+      "loss": 0.0041,
+      "step": 21466
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1787175385628982e-05,
+      "loss": 0.0043,
+      "step": 21467
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1779643950319429e-05,
+      "loss": 0.0008,
+      "step": 21468
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1772114823550665e-05,
+      "loss": 0.0001,
+      "step": 21469
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1764588005448578e-05,
+      "loss": 0.0001,
+      "step": 21470
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.175706349613878e-05,
+      "loss": 0.0004,
+      "step": 21471
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1749541295746973e-05,
+      "loss": 0.0004,
+      "step": 21472
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1742021404398789e-05,
+      "loss": 0.0005,
+      "step": 21473
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1734503822219826e-05,
+      "loss": 0.0004,
+      "step": 21474
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1726988549335636e-05,
+      "loss": 0.0037,
+      "step": 21475
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1719475585871768e-05,
+      "loss": 0.0001,
+      "step": 21476
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.171196493195367e-05,
+      "loss": 0.0011,
+      "step": 21477
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1704456587706794e-05,
+      "loss": 0.0006,
+      "step": 21478
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1696950553256556e-05,
+      "loss": 0.0021,
+      "step": 21479
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1689446828728289e-05,
+      "loss": 0.0044,
+      "step": 21480
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1681945414247374e-05,
+      "loss": 0.0072,
+      "step": 21481
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1674446309939012e-05,
+      "loss": 0.0002,
+      "step": 21482
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1666949515928553e-05,
+      "loss": 0.0048,
+      "step": 21483
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.165945503234113e-05,
+      "loss": 0.0006,
+      "step": 21484
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1651962859301961e-05,
+      "loss": 0.002,
+      "step": 21485
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1644472996936144e-05,
+      "loss": 0.0059,
+      "step": 21486
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1636985445368814e-05,
+      "loss": 0.0083,
+      "step": 21487
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1629500204724984e-05,
+      "loss": 0.0001,
+      "step": 21488
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1622017275129708e-05,
+      "loss": 0.0018,
+      "step": 21489
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1614536656707934e-05,
+      "loss": 0.002,
+      "step": 21490
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.160705834958463e-05,
+      "loss": 0.0002,
+      "step": 21491
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1599582353884696e-05,
+      "loss": 0.0025,
+      "step": 21492
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.159210866973298e-05,
+      "loss": 0.0013,
+      "step": 21493
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1584637297254334e-05,
+      "loss": 0.0008,
+      "step": 21494
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1577168236573475e-05,
+      "loss": 0.0016,
+      "step": 21495
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1569701487815236e-05,
+      "loss": 0.003,
+      "step": 21496
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1562237051104251e-05,
+      "loss": 0.0007,
+      "step": 21497
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1554774926565252e-05,
+      "loss": 0.0014,
+      "step": 21498
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1547315114322875e-05,
+      "loss": 0.0023,
+      "step": 21499
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1539857614501619e-05,
+      "loss": 0.001,
+      "step": 21500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1532402427226167e-05,
+      "loss": 0.0012,
+      "step": 21501
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1524949552620938e-05,
+      "loss": 0.0002,
+      "step": 21502
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1517498990810431e-05,
+      "loss": 0.0033,
+      "step": 21503
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1510050741919113e-05,
+      "loss": 0.0024,
+      "step": 21504
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1502604806071369e-05,
+      "loss": 0.0013,
+      "step": 21505
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1495161183391532e-05,
+      "loss": 0.0001,
+      "step": 21506
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.148771987400397e-05,
+      "loss": 0.002,
+      "step": 21507
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.148028087803295e-05,
+      "loss": 0.0001,
+      "step": 21508
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1472844195602705e-05,
+      "loss": 0.0017,
+      "step": 21509
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1465409826837452e-05,
+      "loss": 0.0014,
+      "step": 21510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.145797777186136e-05,
+      "loss": 0.0005,
+      "step": 21511
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1450548030798561e-05,
+      "loss": 0.0008,
+      "step": 21512
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.144312060377311e-05,
+      "loss": 0.0009,
+      "step": 21513
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1435695490909137e-05,
+      "loss": 0.0028,
+      "step": 21514
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1428272692330559e-05,
+      "loss": 0.0025,
+      "step": 21515
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1420852208161447e-05,
+      "loss": 0.0012,
+      "step": 21516
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1413434038525665e-05,
+      "loss": 0.0008,
+      "step": 21517
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.140601818354715e-05,
+      "loss": 0.0031,
+      "step": 21518
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1398604643349718e-05,
+      "loss": 0.003,
+      "step": 21519
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.139119341805722e-05,
+      "loss": 0.002,
+      "step": 21520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1383784507793459e-05,
+      "loss": 0.0018,
+      "step": 21521
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.137637791268215e-05,
+      "loss": 0.0001,
+      "step": 21522
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1368973632846979e-05,
+      "loss": 0.0026,
+      "step": 21523
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1361571668411646e-05,
+      "loss": 0.0029,
+      "step": 21524
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1354172019499768e-05,
+      "loss": 0.0003,
+      "step": 21525
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1346774686234933e-05,
+      "loss": 0.0076,
+      "step": 21526
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1339379668740672e-05,
+      "loss": 0.0012,
+      "step": 21527
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1331986967140521e-05,
+      "loss": 0.0051,
+      "step": 21528
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1324596581557931e-05,
+      "loss": 0.0023,
+      "step": 21529
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1317208512116388e-05,
+      "loss": 0.0039,
+      "step": 21530
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1309822758939191e-05,
+      "loss": 0.0003,
+      "step": 21531
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1302439322149792e-05,
+      "loss": 0.004,
+      "step": 21532
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1295058201871426e-05,
+      "loss": 0.0018,
+      "step": 21533
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1287679398227446e-05,
+      "loss": 0.001,
+      "step": 21534
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1280302911341049e-05,
+      "loss": 0.0022,
+      "step": 21535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1272928741335441e-05,
+      "loss": 0.0009,
+      "step": 21536
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1265556888333788e-05,
+      "loss": 0.0011,
+      "step": 21537
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1258187352459225e-05,
+      "loss": 0.0005,
+      "step": 21538
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1250820133834804e-05,
+      "loss": 0.0005,
+      "step": 21539
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.124345523258361e-05,
+      "loss": 0.0002,
+      "step": 21540
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1236092648828642e-05,
+      "loss": 0.0013,
+      "step": 21541
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.122873238269284e-05,
+      "loss": 0.0025,
+      "step": 21542
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1221374434299169e-05,
+      "loss": 0.0008,
+      "step": 21543
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1214018803770513e-05,
+      "loss": 0.0024,
+      "step": 21544
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1206665491229744e-05,
+      "loss": 0.0063,
+      "step": 21545
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1199314496799594e-05,
+      "loss": 0.001,
+      "step": 21546
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.119196582060295e-05,
+      "loss": 0.0028,
+      "step": 21547
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1184619462762446e-05,
+      "loss": 0.0032,
+      "step": 21548
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1177275423400883e-05,
+      "loss": 0.0037,
+      "step": 21549
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1169933702640848e-05,
+      "loss": 0.0061,
+      "step": 21550
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1162594300604943e-05,
+      "loss": 0.0009,
+      "step": 21551
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1155257217415836e-05,
+      "loss": 0.002,
+      "step": 21552
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1147922453196012e-05,
+      "loss": 0.0003,
+      "step": 21553
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1140590008067973e-05,
+      "loss": 0.0009,
+      "step": 21554
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1133259882154205e-05,
+      "loss": 0.0018,
+      "step": 21555
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.112593207557711e-05,
+      "loss": 0.0001,
+      "step": 21556
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1118606588459107e-05,
+      "loss": 0.0005,
+      "step": 21557
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1111283420922528e-05,
+      "loss": 0.0001,
+      "step": 21558
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.110396257308968e-05,
+      "loss": 0.0002,
+      "step": 21559
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1096644045082848e-05,
+      "loss": 0.0024,
+      "step": 21560
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1089327837024265e-05,
+      "loss": 0.009,
+      "step": 21561
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1082013949036116e-05,
+      "loss": 0.0035,
+      "step": 21562
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1074702381240574e-05,
+      "loss": 0.0002,
+      "step": 21563
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1067393133759705e-05,
+      "loss": 0.0113,
+      "step": 21564
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1060086206715663e-05,
+      "loss": 0.002,
+      "step": 21565
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1052781600230415e-05,
+      "loss": 0.002,
+      "step": 21566
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1045479314426032e-05,
+      "loss": 0.0005,
+      "step": 21567
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1038179349424431e-05,
+      "loss": 0.0059,
+      "step": 21568
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1030881705347533e-05,
+      "loss": 0.001,
+      "step": 21569
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1023586382317224e-05,
+      "loss": 0.0025,
+      "step": 21570
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1016293380455371e-05,
+      "loss": 0.0004,
+      "step": 21571
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1009002699883763e-05,
+      "loss": 0.0002,
+      "step": 21572
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1001714340724182e-05,
+      "loss": 0.0001,
+      "step": 21573
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0994428303098335e-05,
+      "loss": 0.0032,
+      "step": 21574
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0987144587127939e-05,
+      "loss": 0.0025,
+      "step": 21575
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0979863192934663e-05,
+      "loss": 0.0004,
+      "step": 21576
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0972584120640026e-05,
+      "loss": 0.0011,
+      "step": 21577
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0965307370365733e-05,
+      "loss": 0.0005,
+      "step": 21578
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0958032942233186e-05,
+      "loss": 0.0001,
+      "step": 21579
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0950760836363987e-05,
+      "loss": 0.0003,
+      "step": 21580
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0943491052879587e-05,
+      "loss": 0.0021,
+      "step": 21581
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0936223591901311e-05,
+      "loss": 0.003,
+      "step": 21582
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0928958453550658e-05,
+      "loss": 0.0001,
+      "step": 21583
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0921695637948863e-05,
+      "loss": 0.0021,
+      "step": 21584
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0914435145217332e-05,
+      "loss": 0.0003,
+      "step": 21585
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.090717697547725e-05,
+      "loss": 0.0017,
+      "step": 21586
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0899921128849854e-05,
+      "loss": 0.0002,
+      "step": 21587
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0892667605456362e-05,
+      "loss": 0.0103,
+      "step": 21588
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0885416405417897e-05,
+      "loss": 0.0042,
+      "step": 21589
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0878167528855592e-05,
+      "loss": 0.001,
+      "step": 21590
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0870920975890485e-05,
+      "loss": 0.0002,
+      "step": 21591
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0863676746643629e-05,
+      "loss": 0.001,
+      "step": 21592
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0856434841235994e-05,
+      "loss": 0.0001,
+      "step": 21593
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0849195259788568e-05,
+      "loss": 0.0019,
+      "step": 21594
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0841958002422235e-05,
+      "loss": 0.0027,
+      "step": 21595
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0834723069257916e-05,
+      "loss": 0.0008,
+      "step": 21596
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0827490460416366e-05,
+      "loss": 0.0002,
+      "step": 21597
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0820260176018486e-05,
+      "loss": 0.0001,
+      "step": 21598
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0813032216184913e-05,
+      "loss": 0.0056,
+      "step": 21599
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0805806581036503e-05,
+      "loss": 0.0037,
+      "step": 21600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.079858327069384e-05,
+      "loss": 0.0035,
+      "step": 21601
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0791362285277594e-05,
+      "loss": 0.0047,
+      "step": 21602
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.078414362490837e-05,
+      "loss": 0.0013,
+      "step": 21603
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0776927289706754e-05,
+      "loss": 0.0001,
+      "step": 21604
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0769713279793234e-05,
+      "loss": 0.0002,
+      "step": 21605
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0762501595288332e-05,
+      "loss": 0.0003,
+      "step": 21606
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.075529223631248e-05,
+      "loss": 0.0037,
+      "step": 21607
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0748085202986084e-05,
+      "loss": 0.0001,
+      "step": 21608
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.074088049542955e-05,
+      "loss": 0.0001,
+      "step": 21609
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0733678113763128e-05,
+      "loss": 0.0001,
+      "step": 21610
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0726478058107192e-05,
+      "loss": 0.0011,
+      "step": 21611
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0719280328581976e-05,
+      "loss": 0.0033,
+      "step": 21612
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0712084925307685e-05,
+      "loss": 0.0007,
+      "step": 21613
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0704891848404522e-05,
+      "loss": 0.0003,
+      "step": 21614
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0697701097992561e-05,
+      "loss": 0.0016,
+      "step": 21615
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0690512674191986e-05,
+      "loss": 0.001,
+      "step": 21616
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0683326577122769e-05,
+      "loss": 0.001,
+      "step": 21617
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.067614280690503e-05,
+      "loss": 0.0001,
+      "step": 21618
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0668961363658674e-05,
+      "loss": 0.0031,
+      "step": 21619
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.066178224750367e-05,
+      "loss": 0.0037,
+      "step": 21620
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0654605458559905e-05,
+      "loss": 0.0027,
+      "step": 21621
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0647430996947287e-05,
+      "loss": 0.0013,
+      "step": 21622
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.06402588627856e-05,
+      "loss": 0.002,
+      "step": 21623
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0633089056194649e-05,
+      "loss": 0.0008,
+      "step": 21624
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0625921577294189e-05,
+      "loss": 0.0008,
+      "step": 21625
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.061875642620394e-05,
+      "loss": 0.0001,
+      "step": 21626
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0611593603043573e-05,
+      "loss": 0.001,
+      "step": 21627
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0604433107932658e-05,
+      "loss": 0.0035,
+      "step": 21628
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0597274940990885e-05,
+      "loss": 0.0012,
+      "step": 21629
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0590119102337707e-05,
+      "loss": 0.0033,
+      "step": 21630
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0582965592092747e-05,
+      "loss": 0.0003,
+      "step": 21631
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0575814410375404e-05,
+      "loss": 0.0015,
+      "step": 21632
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0568665557305123e-05,
+      "loss": 0.0001,
+      "step": 21633
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.056151903300137e-05,
+      "loss": 0.0015,
+      "step": 21634
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0554374837583418e-05,
+      "loss": 0.0005,
+      "step": 21635
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0547232971170671e-05,
+      "loss": 0.001,
+      "step": 21636
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0540093433882336e-05,
+      "loss": 0.005,
+      "step": 21637
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0532956225837697e-05,
+      "loss": 0.0015,
+      "step": 21638
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.052582134715596e-05,
+      "loss": 0.0011,
+      "step": 21639
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0518688797956265e-05,
+      "loss": 0.0013,
+      "step": 21640
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0511558578357765e-05,
+      "loss": 0.0021,
+      "step": 21641
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0504430688479548e-05,
+      "loss": 0.0037,
+      "step": 21642
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0497305128440653e-05,
+      "loss": 0.0006,
+      "step": 21643
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0490181898360084e-05,
+      "loss": 0.0013,
+      "step": 21644
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0483060998356813e-05,
+      "loss": 0.0002,
+      "step": 21645
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0475942428549793e-05,
+      "loss": 0.0008,
+      "step": 21646
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0468826189057933e-05,
+      "loss": 0.0021,
+      "step": 21647
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0461712280000001e-05,
+      "loss": 0.0,
+      "step": 21648
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.045460070149492e-05,
+      "loss": 0.0033,
+      "step": 21649
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0447491453661394e-05,
+      "loss": 0.0001,
+      "step": 21650
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0440384536618197e-05,
+      "loss": 0.0038,
+      "step": 21651
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0433279950484013e-05,
+      "loss": 0.0024,
+      "step": 21652
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0426177695377486e-05,
+      "loss": 0.0001,
+      "step": 21653
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0419077771417266e-05,
+      "loss": 0.0044,
+      "step": 21654
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0411980178721929e-05,
+      "loss": 0.0004,
+      "step": 21655
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0404884917410012e-05,
+      "loss": 0.0021,
+      "step": 21656
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0397791987600017e-05,
+      "loss": 0.0025,
+      "step": 21657
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0390701389410422e-05,
+      "loss": 0.0014,
+      "step": 21658
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0383613122959643e-05,
+      "loss": 0.001,
+      "step": 21659
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.037652718836609e-05,
+      "loss": 0.0064,
+      "step": 21660
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0369443585748049e-05,
+      "loss": 0.0002,
+      "step": 21661
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.036236231522391e-05,
+      "loss": 0.0003,
+      "step": 21662
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0355283376911861e-05,
+      "loss": 0.0001,
+      "step": 21663
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0348206770930207e-05,
+      "loss": 0.001,
+      "step": 21664
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0341132497397136e-05,
+      "loss": 0.0036,
+      "step": 21665
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0334060556430723e-05,
+      "loss": 0.0043,
+      "step": 21666
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0326990948149188e-05,
+      "loss": 0.0027,
+      "step": 21667
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0319923672670504e-05,
+      "loss": 0.0094,
+      "step": 21668
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0312858730112811e-05,
+      "loss": 0.0001,
+      "step": 21669
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0305796120594028e-05,
+      "loss": 0.0024,
+      "step": 21670
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.029873584423213e-05,
+      "loss": 0.0001,
+      "step": 21671
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0291677901145073e-05,
+      "loss": 0.0003,
+      "step": 21672
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0284622291450695e-05,
+      "loss": 0.0002,
+      "step": 21673
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0277569015266867e-05,
+      "loss": 0.0023,
+      "step": 21674
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0270518072711365e-05,
+      "loss": 0.0012,
+      "step": 21675
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0263469463901991e-05,
+      "loss": 0.0009,
+      "step": 21676
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0256423188956437e-05,
+      "loss": 0.0006,
+      "step": 21677
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0249379247992424e-05,
+      "loss": 0.0018,
+      "step": 21678
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0242337641127523e-05,
+      "loss": 0.0054,
+      "step": 21679
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.023529836847946e-05,
+      "loss": 0.0004,
+      "step": 21680
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0228261430165674e-05,
+      "loss": 0.0002,
+      "step": 21681
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0221226826303819e-05,
+      "loss": 0.0018,
+      "step": 21682
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0214194557011285e-05,
+      "loss": 0.0056,
+      "step": 21683
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0207164622405579e-05,
+      "loss": 0.0008,
+      "step": 21684
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0200137022604105e-05,
+      "loss": 0.0066,
+      "step": 21685
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0193111757724203e-05,
+      "loss": 0.0003,
+      "step": 21686
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0186088827883299e-05,
+      "loss": 0.0001,
+      "step": 21687
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0179068233198595e-05,
+      "loss": 0.0001,
+      "step": 21688
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0172049973787383e-05,
+      "loss": 0.008,
+      "step": 21689
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0165034049766885e-05,
+      "loss": 0.0002,
+      "step": 21690
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0158020461254258e-05,
+      "loss": 0.0041,
+      "step": 21691
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0151009208366672e-05,
+      "loss": 0.0003,
+      "step": 21692
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0144000291221221e-05,
+      "loss": 0.004,
+      "step": 21693
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.013699370993496e-05,
+      "loss": 0.0016,
+      "step": 21694
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.012998946462491e-05,
+      "loss": 0.0007,
+      "step": 21695
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0122987555408063e-05,
+      "loss": 0.0054,
+      "step": 21696
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0115987982401357e-05,
+      "loss": 0.0004,
+      "step": 21697
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0108990745721734e-05,
+      "loss": 0.0006,
+      "step": 21698
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0101995845485966e-05,
+      "loss": 0.0006,
+      "step": 21699
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0095003281810993e-05,
+      "loss": 0.0036,
+      "step": 21700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0088013054813538e-05,
+      "loss": 0.002,
+      "step": 21701
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0081025164610373e-05,
+      "loss": 0.0032,
+      "step": 21702
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0074039611318192e-05,
+      "loss": 0.0001,
+      "step": 21703
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.006705639505368e-05,
+      "loss": 0.0004,
+      "step": 21704
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.006007551593348e-05,
+      "loss": 0.0026,
+      "step": 21705
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0053096974074165e-05,
+      "loss": 0.0006,
+      "step": 21706
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0046120769592308e-05,
+      "loss": 0.0042,
+      "step": 21707
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.00391469026044e-05,
+      "loss": 0.0019,
+      "step": 21708
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0032175373226947e-05,
+      "loss": 0.0025,
+      "step": 21709
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0025206181576356e-05,
+      "loss": 0.0033,
+      "step": 21710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0018239327769084e-05,
+      "loss": 0.0016,
+      "step": 21711
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0011274811921388e-05,
+      "loss": 0.0014,
+      "step": 21712
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0004312634149708e-05,
+      "loss": 0.0001,
+      "step": 21713
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.997352794570217e-06,
+      "loss": 0.0008,
+      "step": 21714
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.990395293299257e-06,
+      "loss": 0.0033,
+      "step": 21715
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.983440130452968e-06,
+      "loss": 0.0036,
+      "step": 21716
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.97648730614749e-06,
+      "loss": 0.003,
+      "step": 21717
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.969536820499047e-06,
+      "loss": 0.0003,
+      "step": 21718
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.962588673623611e-06,
+      "loss": 0.0043,
+      "step": 21719
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.955642865637326e-06,
+      "loss": 0.0032,
+      "step": 21720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.94869939665613e-06,
+      "loss": 0.0001,
+      "step": 21721
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.941758266796029e-06,
+      "loss": 0.0022,
+      "step": 21722
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.93481947617295e-06,
+      "loss": 0.0004,
+      "step": 21723
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.927883024902766e-06,
+      "loss": 0.0005,
+      "step": 21724
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.920948913101351e-06,
+      "loss": 0.0003,
+      "step": 21725
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.914017140884495e-06,
+      "loss": 0.0028,
+      "step": 21726
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.907087708367989e-06,
+      "loss": 0.0011,
+      "step": 21727
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.900160615667574e-06,
+      "loss": 0.0015,
+      "step": 21728
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.893235862898941e-06,
+      "loss": 0.0012,
+      "step": 21729
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.886313450177696e-06,
+      "loss": 0.0,
+      "step": 21730
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.87939337761955e-06,
+      "loss": 0.0003,
+      "step": 21731
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.872475645339971e-06,
+      "loss": 0.0021,
+      "step": 21732
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.865560253454624e-06,
+      "loss": 0.0018,
+      "step": 21733
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.858647202078913e-06,
+      "loss": 0.0002,
+      "step": 21734
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.851736491328326e-06,
+      "loss": 0.0037,
+      "step": 21735
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.844828121318294e-06,
+      "loss": 0.0008,
+      "step": 21736
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.837922092164186e-06,
+      "loss": 0.0007,
+      "step": 21737
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.831018403981362e-06,
+      "loss": 0.0002,
+      "step": 21738
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.824117056885111e-06,
+      "loss": 0.0013,
+      "step": 21739
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.81721805099071e-06,
+      "loss": 0.0043,
+      "step": 21740
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.810321386413384e-06,
+      "loss": 0.0187,
+      "step": 21741
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.80342706326832e-06,
+      "loss": 0.0012,
+      "step": 21742
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.796535081670648e-06,
+      "loss": 0.0007,
+      "step": 21743
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.789645441735539e-06,
+      "loss": 0.0023,
+      "step": 21744
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.782758143577952e-06,
+      "loss": 0.0027,
+      "step": 21745
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.77587318731301e-06,
+      "loss": 0.0001,
+      "step": 21746
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.768990573055673e-06,
+      "loss": 0.0006,
+      "step": 21747
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.762110300920916e-06,
+      "loss": 0.0014,
+      "step": 21748
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.755232371023647e-06,
+      "loss": 0.0009,
+      "step": 21749
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.748356783478672e-06,
+      "loss": 0.0035,
+      "step": 21750
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.741483538400952e-06,
+      "loss": 0.0005,
+      "step": 21751
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.734612635905175e-06,
+      "loss": 0.0011,
+      "step": 21752
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.727744076106136e-06,
+      "loss": 0.0006,
+      "step": 21753
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.720877859118558e-06,
+      "loss": 0.0,
+      "step": 21754
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.714013985057101e-06,
+      "loss": 0.0009,
+      "step": 21755
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.707152454036437e-06,
+      "loss": 0.0016,
+      "step": 21756
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.700293266171144e-06,
+      "loss": 0.0033,
+      "step": 21757
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.693436421575778e-06,
+      "loss": 0.0026,
+      "step": 21758
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.686581920364883e-06,
+      "loss": 0.0016,
+      "step": 21759
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.679729762652932e-06,
+      "loss": 0.0009,
+      "step": 21760
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.672879948554369e-06,
+      "loss": 0.0026,
+      "step": 21761
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.6660324781836e-06,
+      "loss": 0.0012,
+      "step": 21762
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.659187351654968e-06,
+      "loss": 0.0041,
+      "step": 21763
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.652344569082848e-06,
+      "loss": 0.0004,
+      "step": 21764
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.645504130581449e-06,
+      "loss": 0.0003,
+      "step": 21765
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.63866603626513e-06,
+      "loss": 0.0003,
+      "step": 21766
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.631830286248015e-06,
+      "loss": 0.0002,
+      "step": 21767
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.624996880644281e-06,
+      "loss": 0.0038,
+      "step": 21768
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.618165819568085e-06,
+      "loss": 0.0001,
+      "step": 21769
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.611337103133487e-06,
+      "loss": 0.0007,
+      "step": 21770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.604510731454612e-06,
+      "loss": 0.0008,
+      "step": 21771
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.597686704645402e-06,
+      "loss": 0.0001,
+      "step": 21772
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.59086502281985e-06,
+      "loss": 0.0009,
+      "step": 21773
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.58404568609188e-06,
+      "loss": 0.0015,
+      "step": 21774
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.577228694575433e-06,
+      "loss": 0.0007,
+      "step": 21775
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.57041404838429e-06,
+      "loss": 0.0015,
+      "step": 21776
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.563601747632337e-06,
+      "loss": 0.0004,
+      "step": 21777
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.556791792433322e-06,
+      "loss": 0.0003,
+      "step": 21778
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.549984182901e-06,
+      "loss": 0.0003,
+      "step": 21779
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.543178919149063e-06,
+      "loss": 0.0001,
+      "step": 21780
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.53637600129114e-06,
+      "loss": 0.0003,
+      "step": 21781
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.529575429440923e-06,
+      "loss": 0.0032,
+      "step": 21782
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.522777203711902e-06,
+      "loss": 0.0025,
+      "step": 21783
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.515981324217737e-06,
+      "loss": 0.0012,
+      "step": 21784
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.509187791071821e-06,
+      "loss": 0.0002,
+      "step": 21785
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.502396604387664e-06,
+      "loss": 0.0048,
+      "step": 21786
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.495607764278706e-06,
+      "loss": 0.0027,
+      "step": 21787
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.48882127085831e-06,
+      "loss": 0.0007,
+      "step": 21788
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.482037124239833e-06,
+      "loss": 0.0012,
+      "step": 21789
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.475255324536585e-06,
+      "loss": 0.0004,
+      "step": 21790
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.468475871861825e-06,
+      "loss": 0.0003,
+      "step": 21791
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.461698766328796e-06,
+      "loss": 0.0014,
+      "step": 21792
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.454924008050691e-06,
+      "loss": 0.002,
+      "step": 21793
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.448151597140635e-06,
+      "loss": 0.0004,
+      "step": 21794
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.441381533711806e-06,
+      "loss": 0.0023,
+      "step": 21795
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.434613817877162e-06,
+      "loss": 0.0013,
+      "step": 21796
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.427848449749847e-06,
+      "loss": 0.0002,
+      "step": 21797
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.421085429442803e-06,
+      "loss": 0.0004,
+      "step": 21798
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.414324757068974e-06,
+      "loss": 0.0057,
+      "step": 21799
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.40756643274132e-06,
+      "loss": 0.0003,
+      "step": 21800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.40081045657265e-06,
+      "loss": 0.0044,
+      "step": 21801
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.394056828675905e-06,
+      "loss": 0.0013,
+      "step": 21802
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.387305549163799e-06,
+      "loss": 0.0005,
+      "step": 21803
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.380556618149104e-06,
+      "loss": 0.0008,
+      "step": 21804
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.373810035744567e-06,
+      "loss": 0.0003,
+      "step": 21805
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.367065802062845e-06,
+      "loss": 0.0054,
+      "step": 21806
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.360323917216583e-06,
+      "loss": 0.0001,
+      "step": 21807
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.353584381318391e-06,
+      "loss": 0.0002,
+      "step": 21808
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.346847194480828e-06,
+      "loss": 0.0001,
+      "step": 21809
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.34011235681642e-06,
+      "loss": 0.0005,
+      "step": 21810
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.333379868437663e-06,
+      "loss": 0.0059,
+      "step": 21811
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.32664972945698e-06,
+      "loss": 0.0054,
+      "step": 21812
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.319921939986801e-06,
+      "loss": 0.0005,
+      "step": 21813
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.313196500139436e-06,
+      "loss": 0.0014,
+      "step": 21814
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.30647341002731e-06,
+      "loss": 0.0052,
+      "step": 21815
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.2997526697626e-06,
+      "loss": 0.0001,
+      "step": 21816
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.29303427945765e-06,
+      "loss": 0.0004,
+      "step": 21817
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.286318239224622e-06,
+      "loss": 0.0054,
+      "step": 21818
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.27960454917569e-06,
+      "loss": 0.0018,
+      "step": 21819
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.272893209422999e-06,
+      "loss": 0.0013,
+      "step": 21820
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.26618422007861e-06,
+      "loss": 0.0007,
+      "step": 21821
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.259477581254615e-06,
+      "loss": 0.0008,
+      "step": 21822
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.25277329306301e-06,
+      "loss": 0.0009,
+      "step": 21823
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.24607135561577e-06,
+      "loss": 0.0001,
+      "step": 21824
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.239371769024839e-06,
+      "loss": 0.0002,
+      "step": 21825
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.232674533402113e-06,
+      "loss": 0.0005,
+      "step": 21826
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.2259796488594e-06,
+      "loss": 0.0001,
+      "step": 21827
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.219287115508595e-06,
+      "loss": 0.0038,
+      "step": 21828
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.212596933461392e-06,
+      "loss": 0.0006,
+      "step": 21829
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.2059091028296e-06,
+      "loss": 0.0001,
+      "step": 21830
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.199223623724933e-06,
+      "loss": 0.0034,
+      "step": 21831
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.192540496258948e-06,
+      "loss": 0.0015,
+      "step": 21832
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.185859720543377e-06,
+      "loss": 0.0021,
+      "step": 21833
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.179181296689708e-06,
+      "loss": 0.0,
+      "step": 21834
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.17250522480959e-06,
+      "loss": 0.0083,
+      "step": 21835
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.16583150501443e-06,
+      "loss": 0.0003,
+      "step": 21836
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.159160137415728e-06,
+      "loss": 0.001,
+      "step": 21837
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.152491122124905e-06,
+      "loss": 0.0018,
+      "step": 21838
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.14582445925336e-06,
+      "loss": 0.0028,
+      "step": 21839
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.139160148912417e-06,
+      "loss": 0.0001,
+      "step": 21840
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.132498191213389e-06,
+      "loss": 0.0037,
+      "step": 21841
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.125838586267553e-06,
+      "loss": 0.0063,
+      "step": 21842
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.119181334186138e-06,
+      "loss": 0.0001,
+      "step": 21843
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.112526435080319e-06,
+      "loss": 0.0036,
+      "step": 21844
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.10587388906126e-06,
+      "loss": 0.0056,
+      "step": 21845
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.099223696240072e-06,
+      "loss": 0.0191,
+      "step": 21846
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.09257585672778e-06,
+      "loss": 0.0022,
+      "step": 21847
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.085930370635496e-06,
+      "loss": 0.0016,
+      "step": 21848
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.07928723807415e-06,
+      "loss": 0.0015,
+      "step": 21849
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.07264645915472e-06,
+      "loss": 0.0004,
+      "step": 21850
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.066008033988098e-06,
+      "loss": 0.0005,
+      "step": 21851
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.059371962685197e-06,
+      "loss": 0.0023,
+      "step": 21852
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.052738245356811e-06,
+      "loss": 0.0011,
+      "step": 21853
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.046106882113751e-06,
+      "loss": 0.019,
+      "step": 21854
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.03947787306678e-06,
+      "loss": 0.0091,
+      "step": 21855
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.032851218326609e-06,
+      "loss": 0.0021,
+      "step": 21856
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.026226918003915e-06,
+      "loss": 0.0046,
+      "step": 21857
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.019604972209343e-06,
+      "loss": 0.0008,
+      "step": 21858
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.012985381053506e-06,
+      "loss": 0.0025,
+      "step": 21859
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.006368144646913e-06,
+      "loss": 0.0006,
+      "step": 21860
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.999753263100129e-06,
+      "loss": 0.0003,
+      "step": 21861
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.993140736523613e-06,
+      "loss": 0.0002,
+      "step": 21862
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.986530565027827e-06,
+      "loss": 0.0025,
+      "step": 21863
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.979922748723183e-06,
+      "loss": 0.0021,
+      "step": 21864
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.973317287719978e-06,
+      "loss": 0.005,
+      "step": 21865
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.966714182128621e-06,
+      "loss": 0.0005,
+      "step": 21866
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.960113432059324e-06,
+      "loss": 0.0007,
+      "step": 21867
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.953515037622399e-06,
+      "loss": 0.0029,
+      "step": 21868
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.946918998927993e-06,
+      "loss": 0.0015,
+      "step": 21869
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.9403253160863e-06,
+      "loss": 0.0024,
+      "step": 21870
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.933733989207449e-06,
+      "loss": 0.0001,
+      "step": 21871
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.927145018401516e-06,
+      "loss": 0.0008,
+      "step": 21872
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.92055840377855e-06,
+      "loss": 0.0018,
+      "step": 21873
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.913974145448576e-06,
+      "loss": 0.0032,
+      "step": 21874
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.907392243521544e-06,
+      "loss": 0.0006,
+      "step": 21875
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.900812698107395e-06,
+      "loss": 0.0018,
+      "step": 21876
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.894235509316045e-06,
+      "loss": 0.0047,
+      "step": 21877
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.887660677257252e-06,
+      "loss": 0.0031,
+      "step": 21878
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.881088202040964e-06,
+      "loss": 0.0006,
+      "step": 21879
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.874518083776827e-06,
+      "loss": 0.003,
+      "step": 21880
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.867950322574685e-06,
+      "loss": 0.0017,
+      "step": 21881
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.86138491854415e-06,
+      "loss": 0.0003,
+      "step": 21882
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.854821871794888e-06,
+      "loss": 0.0001,
+      "step": 21883
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.848261182436556e-06,
+      "loss": 0.0003,
+      "step": 21884
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.841702850578686e-06,
+      "loss": 0.0018,
+      "step": 21885
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.835146876330857e-06,
+      "loss": 0.0016,
+      "step": 21886
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.82859325980253e-06,
+      "loss": 0.0017,
+      "step": 21887
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.822042001103185e-06,
+      "loss": 0.0094,
+      "step": 21888
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.81549310034222e-06,
+      "loss": 0.0006,
+      "step": 21889
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.808946557629027e-06,
+      "loss": 0.0002,
+      "step": 21890
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.802402373072937e-06,
+      "loss": 0.0023,
+      "step": 21891
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.795860546783263e-06,
+      "loss": 0.0018,
+      "step": 21892
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.789321078869265e-06,
+      "loss": 0.0001,
+      "step": 21893
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.78278396944016e-06,
+      "loss": 0.0,
+      "step": 21894
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.776249218605124e-06,
+      "loss": 0.0001,
+      "step": 21895
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.769716826473305e-06,
+      "loss": 0.0054,
+      "step": 21896
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.763186793153814e-06,
+      "loss": 0.0016,
+      "step": 21897
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.756659118755683e-06,
+      "loss": 0.004,
+      "step": 21898
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.750133803387987e-06,
+      "loss": 0.0093,
+      "step": 21899
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.743610847159676e-06,
+      "loss": 0.0005,
+      "step": 21900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.737090250179713e-06,
+      "loss": 0.0,
+      "step": 21901
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.730572012556975e-06,
+      "loss": 0.0003,
+      "step": 21902
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.724056134400359e-06,
+      "loss": 0.0022,
+      "step": 21903
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.71754261581868e-06,
+      "loss": 0.0028,
+      "step": 21904
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.71103145692073e-06,
+      "loss": 0.0097,
+      "step": 21905
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.70452265781526e-06,
+      "loss": 0.0039,
+      "step": 21906
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.698016218610965e-06,
+      "loss": 0.0014,
+      "step": 21907
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.69151213941654e-06,
+      "loss": 0.002,
+      "step": 21908
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.685010420340583e-06,
+      "loss": 0.0001,
+      "step": 21909
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.67851106149174e-06,
+      "loss": 0.0006,
+      "step": 21910
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.672014062978471e-06,
+      "loss": 0.0035,
+      "step": 21911
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.665519424909395e-06,
+      "loss": 0.0002,
+      "step": 21912
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.659027147392888e-06,
+      "loss": 0.0003,
+      "step": 21913
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.652537230537449e-06,
+      "loss": 0.0034,
+      "step": 21914
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.646049674451472e-06,
+      "loss": 0.0027,
+      "step": 21915
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.639564479243238e-06,
+      "loss": 0.002,
+      "step": 21916
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.633081645021178e-06,
+      "loss": 0.0022,
+      "step": 21917
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.626601171893438e-06,
+      "loss": 0.0002,
+      "step": 21918
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.620123059968381e-06,
+      "loss": 0.0023,
+      "step": 21919
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.613647309354105e-06,
+      "loss": 0.0033,
+      "step": 21920
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.607173920158805e-06,
+      "loss": 0.0073,
+      "step": 21921
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.600702892490613e-06,
+      "loss": 0.0003,
+      "step": 21922
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.594234226457592e-06,
+      "loss": 0.0085,
+      "step": 21923
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.58776792216777e-06,
+      "loss": 0.0006,
+      "step": 21924
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.581303979729165e-06,
+      "loss": 0.0006,
+      "step": 21925
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.574842399249721e-06,
+      "loss": 0.0039,
+      "step": 21926
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.568383180837368e-06,
+      "loss": 0.0002,
+      "step": 21927
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.561926324600022e-06,
+      "loss": 0.0002,
+      "step": 21928
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.555471830645444e-06,
+      "loss": 0.0023,
+      "step": 21929
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.549019699081517e-06,
+      "loss": 0.0019,
+      "step": 21930
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.542569930015918e-06,
+      "loss": 0.0012,
+      "step": 21931
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.536122523556482e-06,
+      "loss": 0.0049,
+      "step": 21932
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.529677479810803e-06,
+      "loss": 0.0021,
+      "step": 21933
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.523234798886564e-06,
+      "loss": 0.0013,
+      "step": 21934
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.516794480891343e-06,
+      "loss": 0.0045,
+      "step": 21935
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.510356525932705e-06,
+      "loss": 0.003,
+      "step": 21936
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.503920934118247e-06,
+      "loss": 0.0024,
+      "step": 21937
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.497487705555367e-06,
+      "loss": 0.0009,
+      "step": 21938
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.491056840351562e-06,
+      "loss": 0.006,
+      "step": 21939
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.484628338614213e-06,
+      "loss": 0.0003,
+      "step": 21940
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.478202200450701e-06,
+      "loss": 0.0001,
+      "step": 21941
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.471778425968357e-06,
+      "loss": 0.0001,
+      "step": 21942
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.465357015274477e-06,
+      "loss": 0.0002,
+      "step": 21943
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.458937968476277e-06,
+      "loss": 0.0022,
+      "step": 21944
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.452521285681003e-06,
+      "loss": 0.0002,
+      "step": 21945
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.44610696699582e-06,
+      "loss": 0.0017,
+      "step": 21946
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.439695012527842e-06,
+      "loss": 0.0008,
+      "step": 21947
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.4332854223842e-06,
+      "loss": 0.0032,
+      "step": 21948
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.426878196671877e-06,
+      "loss": 0.0002,
+      "step": 21949
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.420473335497968e-06,
+      "loss": 0.0018,
+      "step": 21950
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.414070838969372e-06,
+      "loss": 0.0012,
+      "step": 21951
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.407670707193055e-06,
+      "loss": 0.0029,
+      "step": 21952
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.401272940275927e-06,
+      "loss": 0.0028,
+      "step": 21953
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.394877538324824e-06,
+      "loss": 0.0005,
+      "step": 21954
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.388484501446558e-06,
+      "loss": 0.0002,
+      "step": 21955
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.38209382974791e-06,
+      "loss": 0.0002,
+      "step": 21956
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.375705523335629e-06,
+      "loss": 0.0031,
+      "step": 21957
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.369319582316381e-06,
+      "loss": 0.0001,
+      "step": 21958
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.362936006796861e-06,
+      "loss": 0.002,
+      "step": 21959
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.356554796883668e-06,
+      "loss": 0.004,
+      "step": 21960
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.350175952683385e-06,
+      "loss": 0.0036,
+      "step": 21961
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.343799474302526e-06,
+      "loss": 0.0042,
+      "step": 21962
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.337425361847638e-06,
+      "loss": 0.0002,
+      "step": 21963
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.33105361542512e-06,
+      "loss": 0.0022,
+      "step": 21964
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.324684235141455e-06,
+      "loss": 0.0001,
+      "step": 21965
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.318317221102988e-06,
+      "loss": 0.0015,
+      "step": 21966
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.311952573416036e-06,
+      "loss": 0.0012,
+      "step": 21967
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.305590292186981e-06,
+      "loss": 0.0011,
+      "step": 21968
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.299230377521987e-06,
+      "loss": 0.0038,
+      "step": 21969
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.29287282952737e-06,
+      "loss": 0.0028,
+      "step": 21970
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.286517648309244e-06,
+      "loss": 0.0001,
+      "step": 21971
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.28016483397379e-06,
+      "loss": 0.0003,
+      "step": 21972
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.273814386627076e-06,
+      "loss": 0.0022,
+      "step": 21973
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.267466306375197e-06,
+      "loss": 0.0044,
+      "step": 21974
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.261120593324171e-06,
+      "loss": 0.0053,
+      "step": 21975
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.254777247579997e-06,
+      "loss": 0.0071,
+      "step": 21976
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.248436269248588e-06,
+      "loss": 0.0037,
+      "step": 21977
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.242097658435876e-06,
+      "loss": 0.0006,
+      "step": 21978
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.235761415247743e-06,
+      "loss": 0.0001,
+      "step": 21979
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.22942753978994e-06,
+      "loss": 0.0006,
+      "step": 21980
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.223096032168363e-06,
+      "loss": 0.0004,
+      "step": 21981
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.216766892488646e-06,
+      "loss": 0.0001,
+      "step": 21982
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.21044012085662e-06,
+      "loss": 0.0015,
+      "step": 21983
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.204115717377852e-06,
+      "loss": 0.0002,
+      "step": 21984
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.197793682158022e-06,
+      "loss": 0.0002,
+      "step": 21985
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.191474015302695e-06,
+      "loss": 0.0013,
+      "step": 21986
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.18515671691744e-06,
+      "loss": 0.0004,
+      "step": 21987
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.178841787107771e-06,
+      "loss": 0.0023,
+      "step": 21988
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.172529225979152e-06,
+      "loss": 0.0002,
+      "step": 21989
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.166219033637e-06,
+      "loss": 0.0025,
+      "step": 21990
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.159911210186731e-06,
+      "loss": 0.0041,
+      "step": 21991
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.153605755733678e-06,
+      "loss": 0.0012,
+      "step": 21992
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.147302670383172e-06,
+      "loss": 0.0006,
+      "step": 21993
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.141001954240494e-06,
+      "loss": 0.0038,
+      "step": 21994
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.134703607410815e-06,
+      "loss": 0.0032,
+      "step": 21995
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.128407629999395e-06,
+      "loss": 0.0003,
+      "step": 21996
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.122114022111403e-06,
+      "loss": 0.001,
+      "step": 21997
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.115822783851872e-06,
+      "loss": 0.002,
+      "step": 21998
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.109533915325966e-06,
+      "loss": 0.0015,
+      "step": 21999
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.103247416638636e-06,
+      "loss": 0.0001,
+      "step": 22000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.096963287894965e-06,
+      "loss": 0.0046,
+      "step": 22001
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.09068152919985e-06,
+      "loss": 0.0007,
+      "step": 22002
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.084402140658226e-06,
+      "loss": 0.0064,
+      "step": 22003
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.078125122374973e-06,
+      "loss": 0.002,
+      "step": 22004
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.071850474454927e-06,
+      "loss": 0.0004,
+      "step": 22005
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.065578197002887e-06,
+      "loss": 0.0003,
+      "step": 22006
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.059308290123617e-06,
+      "loss": 0.0005,
+      "step": 22007
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.053040753921819e-06,
+      "loss": 0.001,
+      "step": 22008
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.046775588502192e-06,
+      "loss": 0.001,
+      "step": 22009
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.040512793969367e-06,
+      "loss": 0.0001,
+      "step": 22010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.034252370427946e-06,
+      "loss": 0.0099,
+      "step": 22011
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.027994317982494e-06,
+      "loss": 0.0004,
+      "step": 22012
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.021738636737513e-06,
+      "loss": 0.0021,
+      "step": 22013
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.015485326797532e-06,
+      "loss": 0.0001,
+      "step": 22014
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.009234388266905e-06,
+      "loss": 0.0014,
+      "step": 22015
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.002985821250147e-06,
+      "loss": 0.0003,
+      "step": 22016
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.99673962585154e-06,
+      "loss": 0.0059,
+      "step": 22017
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.990495802175435e-06,
+      "loss": 0.0001,
+      "step": 22018
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.984254350326096e-06,
+      "loss": 0.0001,
+      "step": 22019
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.978015270407779e-06,
+      "loss": 0.0004,
+      "step": 22020
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.971778562524728e-06,
+      "loss": 0.0,
+      "step": 22021
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.965544226781061e-06,
+      "loss": 0.0002,
+      "step": 22022
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.959312263280898e-06,
+      "loss": 0.0064,
+      "step": 22023
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.953082672128353e-06,
+      "loss": 0.0029,
+      "step": 22024
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.946855453427476e-06,
+      "loss": 0.0004,
+      "step": 22025
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.940630607282217e-06,
+      "loss": 0.0016,
+      "step": 22026
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.934408133796611e-06,
+      "loss": 0.0024,
+      "step": 22027
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.92818803307454e-06,
+      "loss": 0.003,
+      "step": 22028
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.921970305219921e-06,
+      "loss": 0.0004,
+      "step": 22029
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.915754950336607e-06,
+      "loss": 0.0026,
+      "step": 22030
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.909541968528343e-06,
+      "loss": 0.0006,
+      "step": 22031
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.903331359898984e-06,
+      "loss": 0.0003,
+      "step": 22032
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.897123124552196e-06,
+      "loss": 0.0001,
+      "step": 22033
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.890917262591712e-06,
+      "loss": 0.0016,
+      "step": 22034
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.884713774121132e-06,
+      "loss": 0.003,
+      "step": 22035
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.878512659244107e-06,
+      "loss": 0.0014,
+      "step": 22036
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.872313918064188e-06,
+      "loss": 0.0006,
+      "step": 22037
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.866117550684926e-06,
+      "loss": 0.0004,
+      "step": 22038
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.859923557209784e-06,
+      "loss": 0.0015,
+      "step": 22039
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.85373193774222e-06,
+      "loss": 0.0018,
+      "step": 22040
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.847542692385662e-06,
+      "loss": 0.0023,
+      "step": 22041
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.841355821243479e-06,
+      "loss": 0.0042,
+      "step": 22042
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.835171324418987e-06,
+      "loss": 0.0011,
+      "step": 22043
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.82898920201549e-06,
+      "loss": 0.0003,
+      "step": 22044
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.82280945413627e-06,
+      "loss": 0.0058,
+      "step": 22045
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.81663208088446e-06,
+      "loss": 0.0022,
+      "step": 22046
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.81045708236333e-06,
+      "loss": 0.0008,
+      "step": 22047
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.804284458675946e-06,
+      "loss": 0.0005,
+      "step": 22048
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.79811420992541e-06,
+      "loss": 0.0086,
+      "step": 22049
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.791946336214822e-06,
+      "loss": 0.0044,
+      "step": 22050
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.785780837647132e-06,
+      "loss": 0.0014,
+      "step": 22051
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.779617714325392e-06,
+      "loss": 0.0002,
+      "step": 22052
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.773456966352487e-06,
+      "loss": 0.0031,
+      "step": 22053
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.7672985938313e-06,
+      "loss": 0.0002,
+      "step": 22054
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.761142596864716e-06,
+      "loss": 0.0001,
+      "step": 22055
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.754988975555553e-06,
+      "loss": 0.0002,
+      "step": 22056
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.74883773000658e-06,
+      "loss": 0.0036,
+      "step": 22057
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.742688860320528e-06,
+      "loss": 0.0001,
+      "step": 22058
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.736542366600101e-06,
+      "loss": 0.0005,
+      "step": 22059
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.730398248947966e-06,
+      "loss": 0.0015,
+      "step": 22060
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.724256507466742e-06,
+      "loss": 0.0003,
+      "step": 22061
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.718117142258978e-06,
+      "loss": 0.0044,
+      "step": 22062
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.71198015342726e-06,
+      "loss": 0.0014,
+      "step": 22063
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.705845541074025e-06,
+      "loss": 0.001,
+      "step": 22064
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.699713305301802e-06,
+      "loss": 0.0066,
+      "step": 22065
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.693583446212947e-06,
+      "loss": 0.0006,
+      "step": 22066
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.687455963909912e-06,
+      "loss": 0.0001,
+      "step": 22067
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.68133085849496e-06,
+      "loss": 0.0024,
+      "step": 22068
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.675208130070448e-06,
+      "loss": 0.0002,
+      "step": 22069
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.669087778738609e-06,
+      "loss": 0.0006,
+      "step": 22070
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.662969804601677e-06,
+      "loss": 0.0008,
+      "step": 22071
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.65685420776182e-06,
+      "loss": 0.0003,
+      "step": 22072
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.650740988321191e-06,
+      "loss": 0.0002,
+      "step": 22073
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.644630146381891e-06,
+      "loss": 0.0016,
+      "step": 22074
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.63852168204599e-06,
+      "loss": 0.0012,
+      "step": 22075
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.63241559541552e-06,
+      "loss": 0.0046,
+      "step": 22076
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.626311886592401e-06,
+      "loss": 0.0035,
+      "step": 22077
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.620210555678669e-06,
+      "loss": 0.0001,
+      "step": 22078
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.6141116027761406e-06,
+      "loss": 0.0004,
+      "step": 22079
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.608015027986736e-06,
+      "loss": 0.0003,
+      "step": 22080
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.601920831412289e-06,
+      "loss": 0.0006,
+      "step": 22081
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.5958290131545194e-06,
+      "loss": 0.0008,
+      "step": 22082
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.589739573315262e-06,
+      "loss": 0.0034,
+      "step": 22083
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.5836525119961185e-06,
+      "loss": 0.0022,
+      "step": 22084
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.577567829298859e-06,
+      "loss": 0.0006,
+      "step": 22085
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.571485525325016e-06,
+      "loss": 0.0006,
+      "step": 22086
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.565405600176244e-06,
+      "loss": 0.0004,
+      "step": 22087
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.559328053954044e-06,
+      "loss": 0.0001,
+      "step": 22088
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.553252886759953e-06,
+      "loss": 0.0014,
+      "step": 22089
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.547180098695421e-06,
+      "loss": 0.0008,
+      "step": 22090
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.541109689861885e-06,
+      "loss": 0.0002,
+      "step": 22091
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.535041660360713e-06,
+      "loss": 0.0019,
+      "step": 22092
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.528976010293275e-06,
+      "loss": 0.0016,
+      "step": 22093
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.522912739760872e-06,
+      "loss": 0.0009,
+      "step": 22094
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.516851848864775e-06,
+      "loss": 0.0023,
+      "step": 22095
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.510793337706233e-06,
+      "loss": 0.0063,
+      "step": 22096
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.504737206386352e-06,
+      "loss": 0.0003,
+      "step": 22097
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.498683455006399e-06,
+      "loss": 0.0024,
+      "step": 22098
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.492632083667394e-06,
+      "loss": 0.0006,
+      "step": 22099
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.4865830924704215e-06,
+      "loss": 0.0003,
+      "step": 22100
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.480536481516536e-06,
+      "loss": 0.0003,
+      "step": 22101
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.474492250906705e-06,
+      "loss": 0.0038,
+      "step": 22102
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.4684504007419155e-06,
+      "loss": 0.0001,
+      "step": 22103
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.4624109311230365e-06,
+      "loss": 0.0013,
+      "step": 22104
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.456373842150953e-06,
+      "loss": 0.0006,
+      "step": 22105
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.450339133926486e-06,
+      "loss": 0.0009,
+      "step": 22106
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.444306806550454e-06,
+      "loss": 0.0019,
+      "step": 22107
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.438276860123593e-06,
+      "loss": 0.0048,
+      "step": 22108
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.432249294746589e-06,
+      "loss": 0.0035,
+      "step": 22109
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.426224110520163e-06,
+      "loss": 0.0001,
+      "step": 22110
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.420201307544915e-06,
+      "loss": 0.0016,
+      "step": 22111
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.414180885921434e-06,
+      "loss": 0.0007,
+      "step": 22112
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.408162845750288e-06,
+      "loss": 0.001,
+      "step": 22113
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.402147187132013e-06,
+      "loss": 0.0009,
+      "step": 22114
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.396133910166996e-06,
+      "loss": 0.0011,
+      "step": 22115
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.390123014955773e-06,
+      "loss": 0.0032,
+      "step": 22116
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.384114501598648e-06,
+      "loss": 0.0,
+      "step": 22117
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.378108370196073e-06,
+      "loss": 0.005,
+      "step": 22118
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.372104620848268e-06,
+      "loss": 0.0009,
+      "step": 22119
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.366103253655553e-06,
+      "loss": 0.0053,
+      "step": 22120
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.3601042687181474e-06,
+      "loss": 0.0001,
+      "step": 22121
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.3541076661362555e-06,
+      "loss": 0.0018,
+      "step": 22122
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.348113446010029e-06,
+      "loss": 0.0004,
+      "step": 22123
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.342121608439588e-06,
+      "loss": 0.0014,
+      "step": 22124
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.336132153525004e-06,
+      "loss": 0.0007,
+      "step": 22125
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.330145081366295e-06,
+      "loss": 0.0032,
+      "step": 22126
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.324160392063499e-06,
+      "loss": 0.0016,
+      "step": 22127
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.318178085716519e-06,
+      "loss": 0.0003,
+      "step": 22128
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.312198162425326e-06,
+      "loss": 0.0017,
+      "step": 22129
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.306220622289721e-06,
+      "loss": 0.0018,
+      "step": 22130
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.300245465409627e-06,
+      "loss": 0.0015,
+      "step": 22131
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.294272691884778e-06,
+      "loss": 0.0028,
+      "step": 22132
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.28830230181493e-06,
+      "loss": 0.0005,
+      "step": 22133
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.282334295299852e-06,
+      "loss": 0.0008,
+      "step": 22134
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.276368672439165e-06,
+      "loss": 0.0037,
+      "step": 22135
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.270405433332571e-06,
+      "loss": 0.0027,
+      "step": 22136
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.264444578079609e-06,
+      "loss": 0.0027,
+      "step": 22137
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.258486106779848e-06,
+      "loss": 0.0003,
+      "step": 22138
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.252530019532809e-06,
+      "loss": 0.0043,
+      "step": 22139
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.24657631643798e-06,
+      "loss": 0.0002,
+      "step": 22140
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.240624997594796e-06,
+      "loss": 0.0019,
+      "step": 22141
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.234676063102646e-06,
+      "loss": 0.0037,
+      "step": 22142
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.228729513060899e-06,
+      "loss": 0.0009,
+      "step": 22143
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.2227853475688606e-06,
+      "loss": 0.0019,
+      "step": 22144
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.21684356672585e-06,
+      "loss": 0.0003,
+      "step": 22145
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.210904170631021e-06,
+      "loss": 0.0003,
+      "step": 22146
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.204967159383679e-06,
+      "loss": 0.0004,
+      "step": 22147
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.199032533082877e-06,
+      "loss": 0.0001,
+      "step": 22148
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.193100291827836e-06,
+      "loss": 0.0001,
+      "step": 22149
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.18717043571756e-06,
+      "loss": 0.0053,
+      "step": 22150
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.181242964851119e-06,
+      "loss": 0.0009,
+      "step": 22151
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.175317879327519e-06,
+      "loss": 0.0016,
+      "step": 22152
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.169395179245696e-06,
+      "loss": 0.0009,
+      "step": 22153
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.163474864704605e-06,
+      "loss": 0.0015,
+      "step": 22154
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.157556935803099e-06,
+      "loss": 0.0032,
+      "step": 22155
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.151641392640018e-06,
+      "loss": 0.0019,
+      "step": 22156
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.14572823531418e-06,
+      "loss": 0.0009,
+      "step": 22157
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.139817463924324e-06,
+      "loss": 0.0003,
+      "step": 22158
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.133909078569189e-06,
+      "loss": 0.0007,
+      "step": 22159
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.1280030793474775e-06,
+      "loss": 0.0,
+      "step": 22160
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.122099466357762e-06,
+      "loss": 0.0054,
+      "step": 22161
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.116198239698711e-06,
+      "loss": 0.0004,
+      "step": 22162
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.110299399468866e-06,
+      "loss": 0.0004,
+      "step": 22163
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.104402945766729e-06,
+      "loss": 0.0008,
+      "step": 22164
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.098508878690823e-06,
+      "loss": 0.0005,
+      "step": 22165
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.092617198339534e-06,
+      "loss": 0.0059,
+      "step": 22166
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.086727904811318e-06,
+      "loss": 0.0003,
+      "step": 22167
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.080840998204479e-06,
+      "loss": 0.0003,
+      "step": 22168
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.074956478617422e-06,
+      "loss": 0.0001,
+      "step": 22169
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.069074346148368e-06,
+      "loss": 0.0035,
+      "step": 22170
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.0631946008955555e-06,
+      "loss": 0.0002,
+      "step": 22171
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.0573172429572224e-06,
+      "loss": 0.0007,
+      "step": 22172
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.05144227243149e-06,
+      "loss": 0.0015,
+      "step": 22173
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.0455696894165306e-06,
+      "loss": 0.0041,
+      "step": 22174
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.039699494010398e-06,
+      "loss": 0.0003,
+      "step": 22175
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.033831686311131e-06,
+      "loss": 0.0,
+      "step": 22176
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.027966266416751e-06,
+      "loss": 0.001,
+      "step": 22177
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.022103234425231e-06,
+      "loss": 0.0008,
+      "step": 22178
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.01624259043444e-06,
+      "loss": 0.003,
+      "step": 22179
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.010384334542335e-06,
+      "loss": 0.0039,
+      "step": 22180
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.0045284668466875e-06,
+      "loss": 0.0012,
+      "step": 22181
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.998674987445369e-06,
+      "loss": 0.0003,
+      "step": 22182
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.9928238964361005e-06,
+      "loss": 0.0045,
+      "step": 22183
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.986975193916605e-06,
+      "loss": 0.0019,
+      "step": 22184
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.981128879984588e-06,
+      "loss": 0.0004,
+      "step": 22185
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.97528495473767e-06,
+      "loss": 0.0024,
+      "step": 22186
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.969443418273507e-06,
+      "loss": 0.0038,
+      "step": 22187
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.963604270689621e-06,
+      "loss": 0.0001,
+      "step": 22188
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.957767512083534e-06,
+      "loss": 0.0032,
+      "step": 22189
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.951933142552734e-06,
+      "loss": 0.0002,
+      "step": 22190
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.946101162194678e-06,
+      "loss": 0.0004,
+      "step": 22191
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.94027157110677e-06,
+      "loss": 0.0026,
+      "step": 22192
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.934444369386366e-06,
+      "loss": 0.0001,
+      "step": 22193
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.9286195571308045e-06,
+      "loss": 0.0001,
+      "step": 22194
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.922797134437358e-06,
+      "loss": 0.0013,
+      "step": 22195
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.9169771014032985e-06,
+      "loss": 0.0006,
+      "step": 22196
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.911159458125764e-06,
+      "loss": 0.0001,
+      "step": 22197
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.905344204702012e-06,
+      "loss": 0.0018,
+      "step": 22198
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.899531341229081e-06,
+      "loss": 0.0026,
+      "step": 22199
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.893720867804126e-06,
+      "loss": 0.001,
+      "step": 22200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.887912784524169e-06,
+      "loss": 0.001,
+      "step": 22201
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.8821070914861835e-06,
+      "loss": 0.0003,
+      "step": 22202
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.87630378878719e-06,
+      "loss": 0.0002,
+      "step": 22203
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.870502876524065e-06,
+      "loss": 0.0002,
+      "step": 22204
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.864704354793727e-06,
+      "loss": 0.0013,
+      "step": 22205
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.858908223693016e-06,
+      "loss": 0.0029,
+      "step": 22206
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.853114483318739e-06,
+      "loss": 0.0003,
+      "step": 22207
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.847323133767651e-06,
+      "loss": 0.006,
+      "step": 22208
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.841534175136493e-06,
+      "loss": 0.0007,
+      "step": 22209
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.835747607521952e-06,
+      "loss": 0.0008,
+      "step": 22210
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.829963431020668e-06,
+      "loss": 0.0004,
+      "step": 22211
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.824181645729215e-06,
+      "loss": 0.0002,
+      "step": 22212
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.8184022517442485e-06,
+      "loss": 0.0004,
+      "step": 22213
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.812625249162173e-06,
+      "loss": 0.0013,
+      "step": 22214
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.80685063807958e-06,
+      "loss": 0.0013,
+      "step": 22215
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.801078418592892e-06,
+      "loss": 0.0006,
+      "step": 22216
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.795308590798465e-06,
+      "loss": 0.0063,
+      "step": 22217
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.789541154792721e-06,
+      "loss": 0.0016,
+      "step": 22218
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.783776110671968e-06,
+      "loss": 0.0002,
+      "step": 22219
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.7780134585324945e-06,
+      "loss": 0.0005,
+      "step": 22220
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.77225319847054e-06,
+      "loss": 0.0003,
+      "step": 22221
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.766495330582328e-06,
+      "loss": 0.0003,
+      "step": 22222
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.760739854964014e-06,
+      "loss": 0.0005,
+      "step": 22223
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.754986771711724e-06,
+      "loss": 0.0046,
+      "step": 22224
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.749236080921544e-06,
+      "loss": 0.003,
+      "step": 22225
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.743487782689533e-06,
+      "loss": 0.0001,
+      "step": 22226
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.737741877111697e-06,
+      "loss": 0.0081,
+      "step": 22227
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.731998364283991e-06,
+      "loss": 0.0,
+      "step": 22228
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.7262572443023575e-06,
+      "loss": 0.002,
+      "step": 22229
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.720518517262652e-06,
+      "loss": 0.0008,
+      "step": 22230
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.714782183260781e-06,
+      "loss": 0.0001,
+      "step": 22231
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.709048242392484e-06,
+      "loss": 0.0001,
+      "step": 22232
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.703316694753602e-06,
+      "loss": 0.0005,
+      "step": 22233
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6975875404397915e-06,
+      "loss": 0.001,
+      "step": 22234
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.691860779546792e-06,
+      "loss": 0.0116,
+      "step": 22235
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.686136412170212e-06,
+      "loss": 0.0045,
+      "step": 22236
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.68041443840569e-06,
+      "loss": 0.0001,
+      "step": 22237
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.674694858348767e-06,
+      "loss": 0.003,
+      "step": 22238
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.668977672094999e-06,
+      "loss": 0.0069,
+      "step": 22239
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.663262879739861e-06,
+      "loss": 0.0005,
+      "step": 22240
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.657550481378793e-06,
+      "loss": 0.0006,
+      "step": 22241
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.651840477107217e-06,
+      "loss": 0.0001,
+      "step": 22242
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.646132867020476e-06,
+      "loss": 0.0043,
+      "step": 22243
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6404276512139584e-06,
+      "loss": 0.001,
+      "step": 22244
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.634724829782856e-06,
+      "loss": 0.0149,
+      "step": 22245
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.629024402822492e-06,
+      "loss": 0.0006,
+      "step": 22246
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.623326370428072e-06,
+      "loss": 0.0022,
+      "step": 22247
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.617630732694707e-06,
+      "loss": 0.0001,
+      "step": 22248
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6119374897176005e-06,
+      "loss": 0.0001,
+      "step": 22249
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.606246641591761e-06,
+      "loss": 0.0009,
+      "step": 22250
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.60055818841233e-06,
+      "loss": 0.0021,
+      "step": 22251
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.594872130274232e-06,
+      "loss": 0.0021,
+      "step": 22252
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.589188467272455e-06,
+      "loss": 0.0029,
+      "step": 22253
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.583507199501959e-06,
+      "loss": 0.004,
+      "step": 22254
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.577828327057599e-06,
+      "loss": 0.0007,
+      "step": 22255
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.572151850034235e-06,
+      "loss": 0.0056,
+      "step": 22256
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.566477768526674e-06,
+      "loss": 0.0001,
+      "step": 22257
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.560806082629688e-06,
+      "loss": 0.002,
+      "step": 22258
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.555136792438004e-06,
+      "loss": 0.003,
+      "step": 22259
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.549469898046294e-06,
+      "loss": 0.0001,
+      "step": 22260
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5438053995492176e-06,
+      "loss": 0.0015,
+      "step": 22261
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.538143297041415e-06,
+      "loss": 0.004,
+      "step": 22262
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5324835906173935e-06,
+      "loss": 0.0004,
+      "step": 22263
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5268262803717276e-06,
+      "loss": 0.0002,
+      "step": 22264
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5211713663988585e-06,
+      "loss": 0.0013,
+      "step": 22265
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.515518848793294e-06,
+      "loss": 0.001,
+      "step": 22266
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.509868727649409e-06,
+      "loss": 0.0001,
+      "step": 22267
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.504221003061544e-06,
+      "loss": 0.0038,
+      "step": 22268
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.49857567512409e-06,
+      "loss": 0.0038,
+      "step": 22269
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.49293274393129e-06,
+      "loss": 0.0003,
+      "step": 22270
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.487292209577383e-06,
+      "loss": 0.0137,
+      "step": 22271
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.481654072156611e-06,
+      "loss": 0.0009,
+      "step": 22272
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.4760183317631335e-06,
+      "loss": 0.003,
+      "step": 22273
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.4703849884910564e-06,
+      "loss": 0.0055,
+      "step": 22274
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.464754042434505e-06,
+      "loss": 0.0,
+      "step": 22275
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.459125493687473e-06,
+      "loss": 0.0003,
+      "step": 22276
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.453499342343998e-06,
+      "loss": 0.0019,
+      "step": 22277
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.4478755884980584e-06,
+      "loss": 0.0026,
+      "step": 22278
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.442254232243576e-06,
+      "loss": 0.0005,
+      "step": 22279
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.436635273674446e-06,
+      "loss": 0.006,
+      "step": 22280
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.431018712884456e-06,
+      "loss": 0.002,
+      "step": 22281
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.425404549967516e-06,
+      "loss": 0.0023,
+      "step": 22282
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.419792785017286e-06,
+      "loss": 0.0031,
+      "step": 22283
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.414183418127572e-06,
+      "loss": 0.002,
+      "step": 22284
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.4085764493920335e-06,
+      "loss": 0.0028,
+      "step": 22285
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.402971878904295e-06,
+      "loss": 0.0004,
+      "step": 22286
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.397369706757999e-06,
+      "loss": 0.0008,
+      "step": 22287
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.391769933046703e-06,
+      "loss": 0.0012,
+      "step": 22288
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.386172557863917e-06,
+      "loss": 0.0001,
+      "step": 22289
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.380577581303148e-06,
+      "loss": 0.0016,
+      "step": 22290
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.374985003457822e-06,
+      "loss": 0.0004,
+      "step": 22291
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.369394824421364e-06,
+      "loss": 0.0042,
+      "step": 22292
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.363807044287167e-06,
+      "loss": 0.0002,
+      "step": 22293
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.358221663148472e-06,
+      "loss": 0.0091,
+      "step": 22294
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.352638681098654e-06,
+      "loss": 0.0004,
+      "step": 22295
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.347058098230889e-06,
+      "loss": 0.0001,
+      "step": 22296
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.341479914638454e-06,
+      "loss": 0.0002,
+      "step": 22297
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.335904130414471e-06,
+      "loss": 0.001,
+      "step": 22298
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.330330745652051e-06,
+      "loss": 0.0015,
+      "step": 22299
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.324759760444337e-06,
+      "loss": 0.0003,
+      "step": 22300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.319191174884302e-06,
+      "loss": 0.0015,
+      "step": 22301
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.313624989065041e-06,
+      "loss": 0.0012,
+      "step": 22302
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.308061203079429e-06,
+      "loss": 0.0034,
+      "step": 22303
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.3024998170204566e-06,
+      "loss": 0.0008,
+      "step": 22304
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.296940830980967e-06,
+      "loss": 0.0032,
+      "step": 22305
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.291384245053837e-06,
+      "loss": 0.0007,
+      "step": 22306
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.285830059331842e-06,
+      "loss": 0.0004,
+      "step": 22307
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.280278273907791e-06,
+      "loss": 0.0017,
+      "step": 22308
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.274728888874358e-06,
+      "loss": 0.0017,
+      "step": 22309
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.269181904324272e-06,
+      "loss": 0.0102,
+      "step": 22310
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.26363732035014e-06,
+      "loss": 0.0023,
+      "step": 22311
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.2580951370446055e-06,
+      "loss": 0.0038,
+      "step": 22312
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.252555354500227e-06,
+      "loss": 0.0007,
+      "step": 22313
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.2470179728094815e-06,
+      "loss": 0.0052,
+      "step": 22314
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.241482992064911e-06,
+      "loss": 0.0004,
+      "step": 22315
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.235950412358908e-06,
+      "loss": 0.0006,
+      "step": 22316
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.230420233783951e-06,
+      "loss": 0.0049,
+      "step": 22317
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.224892456432345e-06,
+      "loss": 0.002,
+      "step": 22318
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.21936708039642e-06,
+      "loss": 0.0004,
+      "step": 22319
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.213844105768467e-06,
+      "loss": 0.0017,
+      "step": 22320
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.208323532640746e-06,
+      "loss": 0.0032,
+      "step": 22321
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.202805361105451e-06,
+      "loss": 0.0017,
+      "step": 22322
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.197289591254722e-06,
+      "loss": 0.0011,
+      "step": 22323
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.1917762231807225e-06,
+      "loss": 0.0041,
+      "step": 22324
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.186265256975509e-06,
+      "loss": 0.0031,
+      "step": 22325
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.180756692731142e-06,
+      "loss": 0.0017,
+      "step": 22326
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.1752505305396e-06,
+      "loss": 0.0036,
+      "step": 22327
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.16974677049289e-06,
+      "loss": 0.0047,
+      "step": 22328
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.164245412682872e-06,
+      "loss": 0.0061,
+      "step": 22329
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.158746457201475e-06,
+      "loss": 0.0001,
+      "step": 22330
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.153249904140556e-06,
+      "loss": 0.0014,
+      "step": 22331
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.14775575359186e-06,
+      "loss": 0.0001,
+      "step": 22332
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.142264005647213e-06,
+      "loss": 0.0004,
+      "step": 22333
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.136774660398275e-06,
+      "loss": 0.0003,
+      "step": 22334
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.131287717936806e-06,
+      "loss": 0.0001,
+      "step": 22335
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.125803178354383e-06,
+      "loss": 0.0019,
+      "step": 22336
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.120321041742615e-06,
+      "loss": 0.0051,
+      "step": 22337
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.11484130819308e-06,
+      "loss": 0.0006,
+      "step": 22338
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.109363977797305e-06,
+      "loss": 0.0024,
+      "step": 22339
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.1038890506467656e-06,
+      "loss": 0.0009,
+      "step": 22340
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.098416526832905e-06,
+      "loss": 0.0001,
+      "step": 22341
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.092946406447119e-06,
+      "loss": 0.0001,
+      "step": 22342
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.087478689580766e-06,
+      "loss": 0.0,
+      "step": 22343
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.08201337632519e-06,
+      "loss": 0.003,
+      "step": 22344
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.076550466771634e-06,
+      "loss": 0.0002,
+      "step": 22345
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.071089961011377e-06,
+      "loss": 0.0012,
+      "step": 22346
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.065631859135578e-06,
+      "loss": 0.0011,
+      "step": 22347
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0601761612354636e-06,
+      "loss": 0.0002,
+      "step": 22348
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.054722867402079e-06,
+      "loss": 0.0062,
+      "step": 22349
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0492719777265494e-06,
+      "loss": 0.0005,
+      "step": 22350
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0438234922999044e-06,
+      "loss": 0.0034,
+      "step": 22351
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0383774112131185e-06,
+      "loss": 0.0001,
+      "step": 22352
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0329337345572215e-06,
+      "loss": 0.0006,
+      "step": 22353
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.027492462423056e-06,
+      "loss": 0.0003,
+      "step": 22354
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.022053594901549e-06,
+      "loss": 0.0005,
+      "step": 22355
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.016617132083512e-06,
+      "loss": 0.0001,
+      "step": 22356
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.011183074059772e-06,
+      "loss": 0.001,
+      "step": 22357
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0057514209210566e-06,
+      "loss": 0.0001,
+      "step": 22358
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.00032217275811e-06,
+      "loss": 0.0002,
+      "step": 22359
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.994895329661608e-06,
+      "loss": 0.0004,
+      "step": 22360
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.989470891722165e-06,
+      "loss": 0.0001,
+      "step": 22361
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.984048859030421e-06,
+      "loss": 0.0009,
+      "step": 22362
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.97862923167689e-06,
+      "loss": 0.0043,
+      "step": 22363
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.973212009752132e-06,
+      "loss": 0.0006,
+      "step": 22364
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.967797193346574e-06,
+      "loss": 0.0005,
+      "step": 22365
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.962384782550711e-06,
+      "loss": 0.0021,
+      "step": 22366
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.9569747774549035e-06,
+      "loss": 0.009,
+      "step": 22367
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.951567178149513e-06,
+      "loss": 0.0059,
+      "step": 22368
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.946161984724868e-06,
+      "loss": 0.0024,
+      "step": 22369
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.940759197271228e-06,
+      "loss": 0.0075,
+      "step": 22370
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.9353588158788375e-06,
+      "loss": 0.0037,
+      "step": 22371
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.929960840637909e-06,
+      "loss": 0.0032,
+      "step": 22372
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.924565271638587e-06,
+      "loss": 0.0004,
+      "step": 22373
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.919172108970966e-06,
+      "loss": 0.002,
+      "step": 22374
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.913781352725155e-06,
+      "loss": 0.0004,
+      "step": 22375
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.908393002991185e-06,
+      "loss": 0.0011,
+      "step": 22376
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.903007059859049e-06,
+      "loss": 0.004,
+      "step": 22377
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8976235234186765e-06,
+      "loss": 0.002,
+      "step": 22378
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8922423937600275e-06,
+      "loss": 0.0013,
+      "step": 22379
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.88686367097293e-06,
+      "loss": 0.0017,
+      "step": 22380
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.881487355147263e-06,
+      "loss": 0.0063,
+      "step": 22381
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.876113446372788e-06,
+      "loss": 0.0044,
+      "step": 22382
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.870741944739249e-06,
+      "loss": 0.0048,
+      "step": 22383
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.865372850336425e-06,
+      "loss": 0.0001,
+      "step": 22384
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.860006163253911e-06,
+      "loss": 0.0003,
+      "step": 22385
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.854641883581418e-06,
+      "loss": 0.0004,
+      "step": 22386
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8492800114084745e-06,
+      "loss": 0.0006,
+      "step": 22387
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.84392054682466e-06,
+      "loss": 0.0002,
+      "step": 22388
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.838563489919484e-06,
+      "loss": 0.0006,
+      "step": 22389
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.833208840782427e-06,
+      "loss": 0.0001,
+      "step": 22390
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.827856599502917e-06,
+      "loss": 0.0006,
+      "step": 22391
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8225067661703315e-06,
+      "loss": 0.0009,
+      "step": 22392
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.81715934087405e-06,
+      "loss": 0.0026,
+      "step": 22393
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.811814323703384e-06,
+      "loss": 0.0022,
+      "step": 22394
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.806471714747596e-06,
+      "loss": 0.0028,
+      "step": 22395
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.80113151409588e-06,
+      "loss": 0.0054,
+      "step": 22396
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.7957937218375e-06,
+      "loss": 0.0005,
+      "step": 22397
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.790458338061532e-06,
+      "loss": 0.0017,
+      "step": 22398
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.785125362857157e-06,
+      "loss": 0.0001,
+      "step": 22399
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.7797947963134025e-06,
+      "loss": 0.0008,
+      "step": 22400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.774466638519298e-06,
+      "loss": 0.0007,
+      "step": 22401
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.769140889563855e-06,
+      "loss": 0.0015,
+      "step": 22402
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.763817549536021e-06,
+      "loss": 0.0019,
+      "step": 22403
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.758496618524688e-06,
+      "loss": 0.0002,
+      "step": 22404
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.753178096618738e-06,
+      "loss": 0.0008,
+      "step": 22405
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.747861983906998e-06,
+      "loss": 0.003,
+      "step": 22406
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.742548280478266e-06,
+      "loss": 0.0035,
+      "step": 22407
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.7372369864212846e-06,
+      "loss": 0.0004,
+      "step": 22408
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.731928101824751e-06,
+      "loss": 0.0003,
+      "step": 22409
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.726621626777361e-06,
+      "loss": 0.0001,
+      "step": 22410
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.721317561367694e-06,
+      "loss": 0.0012,
+      "step": 22411
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.716015905684379e-06,
+      "loss": 0.0005,
+      "step": 22412
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.710716659815962e-06,
+      "loss": 0.0002,
+      "step": 22413
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.705419823850938e-06,
+      "loss": 0.0005,
+      "step": 22414
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.700125397877803e-06,
+      "loss": 0.0035,
+      "step": 22415
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.69483338198492e-06,
+      "loss": 0.0002,
+      "step": 22416
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.689543776260752e-06,
+      "loss": 0.0022,
+      "step": 22417
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.684256580793578e-06,
+      "loss": 0.0006,
+      "step": 22418
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.678971795671744e-06,
+      "loss": 0.002,
+      "step": 22419
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.6736894209835126e-06,
+      "loss": 0.0019,
+      "step": 22420
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.6684094568170965e-06,
+      "loss": 0.002,
+      "step": 22421
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.6631319032606924e-06,
+      "loss": 0.0002,
+      "step": 22422
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.657856760402429e-06,
+      "loss": 0.0013,
+      "step": 22423
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.652584028330437e-06,
+      "loss": 0.0014,
+      "step": 22424
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.647313707132761e-06,
+      "loss": 0.0052,
+      "step": 22425
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.642045796897432e-06,
+      "loss": 0.0017,
+      "step": 22426
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.636780297712428e-06,
+      "loss": 0.0014,
+      "step": 22427
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.631517209665715e-06,
+      "loss": 0.0014,
+      "step": 22428
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.626256532845153e-06,
+      "loss": 0.0019,
+      "step": 22429
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.6209982673386554e-06,
+      "loss": 0.004,
+      "step": 22430
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.615742413234004e-06,
+      "loss": 0.0019,
+      "step": 22431
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.6104889706190256e-06,
+      "loss": 0.0022,
+      "step": 22432
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.605237939581403e-06,
+      "loss": 0.0003,
+      "step": 22433
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.599989320208881e-06,
+      "loss": 0.0012,
+      "step": 22434
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.594743112589123e-06,
+      "loss": 0.0003,
+      "step": 22435
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.589499316809709e-06,
+      "loss": 0.0028,
+      "step": 22436
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.584257932958286e-06,
+      "loss": 0.0011,
+      "step": 22437
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.5790189611223504e-06,
+      "loss": 0.0034,
+      "step": 22438
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.573782401389415e-06,
+      "loss": 0.0039,
+      "step": 22439
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.568548253846928e-06,
+      "loss": 0.0001,
+      "step": 22440
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.563316518582334e-06,
+      "loss": 0.0005,
+      "step": 22441
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.558087195682964e-06,
+      "loss": 0.0019,
+      "step": 22442
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.552860285236216e-06,
+      "loss": 0.0038,
+      "step": 22443
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.547635787329352e-06,
+      "loss": 0.0007,
+      "step": 22444
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.542413702049653e-06,
+      "loss": 0.0003,
+      "step": 22445
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.537194029484349e-06,
+      "loss": 0.0009,
+      "step": 22446
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.531976769720553e-06,
+      "loss": 0.0003,
+      "step": 22447
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.526761922845496e-06,
+      "loss": 0.0027,
+      "step": 22448
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.521549488946175e-06,
+      "loss": 0.0022,
+      "step": 22449
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.516339468109754e-06,
+      "loss": 0.0,
+      "step": 22450
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.511131860423163e-06,
+      "loss": 0.0013,
+      "step": 22451
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.505926665973415e-06,
+      "loss": 0.001,
+      "step": 22452
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.500723884847458e-06,
+      "loss": 0.0001,
+      "step": 22453
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.495523517132156e-06,
+      "loss": 0.0007,
+      "step": 22454
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.490325562914388e-06,
+      "loss": 0.0009,
+      "step": 22455
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.48513002228097e-06,
+      "loss": 0.0016,
+      "step": 22456
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.479936895318665e-06,
+      "loss": 0.0008,
+      "step": 22457
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.47474618211422e-06,
+      "loss": 0.0008,
+      "step": 22458
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.469557882754316e-06,
+      "loss": 0.0002,
+      "step": 22459
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.464371997325617e-06,
+      "loss": 0.0016,
+      "step": 22460
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.459188525914754e-06,
+      "loss": 0.0001,
+      "step": 22461
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.454007468608257e-06,
+      "loss": 0.0021,
+      "step": 22462
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.448828825492723e-06,
+      "loss": 0.0003,
+      "step": 22463
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.443652596654552e-06,
+      "loss": 0.0002,
+      "step": 22464
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.438478782180289e-06,
+      "loss": 0.0006,
+      "step": 22465
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.433307382156316e-06,
+      "loss": 0.0074,
+      "step": 22466
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.428138396668963e-06,
+      "loss": 0.0022,
+      "step": 22467
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.422971825804645e-06,
+      "loss": 0.0008,
+      "step": 22468
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.417807669649577e-06,
+      "loss": 0.0005,
+      "step": 22469
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.412645928290038e-06,
+      "loss": 0.0004,
+      "step": 22470
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.407486601812244e-06,
+      "loss": 0.0004,
+      "step": 22471
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.402329690302359e-06,
+      "loss": 0.0007,
+      "step": 22472
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.397175193846531e-06,
+      "loss": 0.0008,
+      "step": 22473
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.392023112530808e-06,
+      "loss": 0.0018,
+      "step": 22474
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.386873446441287e-06,
+      "loss": 0.0001,
+      "step": 22475
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.381726195663949e-06,
+      "loss": 0.0007,
+      "step": 22476
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.376581360284776e-06,
+      "loss": 0.0034,
+      "step": 22477
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.371438940389683e-06,
+      "loss": 0.0016,
+      "step": 22478
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.366298936064584e-06,
+      "loss": 0.0001,
+      "step": 22479
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.3611613473952595e-06,
+      "loss": 0.0002,
+      "step": 22480
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.356026174467626e-06,
+      "loss": 0.0022,
+      "step": 22481
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.350893417367346e-06,
+      "loss": 0.0021,
+      "step": 22482
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.345763076180221e-06,
+      "loss": 0.0003,
+      "step": 22483
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.340635150991879e-06,
+      "loss": 0.0034,
+      "step": 22484
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.33550964188802e-06,
+      "loss": 0.001,
+      "step": 22485
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.330386548954208e-06,
+      "loss": 0.0014,
+      "step": 22486
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.325265872276025e-06,
+      "loss": 0.0011,
+      "step": 22487
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.320147611938985e-06,
+      "loss": 0.0002,
+      "step": 22488
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.315031768028605e-06,
+      "loss": 0.003,
+      "step": 22489
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.309918340630281e-06,
+      "loss": 0.0078,
+      "step": 22490
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.304807329829463e-06,
+      "loss": 0.0014,
+      "step": 22491
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2996987357115e-06,
+      "loss": 0.005,
+      "step": 22492
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.294592558361688e-06,
+      "loss": 0.0002,
+      "step": 22493
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.28948879786536e-06,
+      "loss": 0.0002,
+      "step": 22494
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2843874543076815e-06,
+      "loss": 0.0019,
+      "step": 22495
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.279288527773934e-06,
+      "loss": 0.0015,
+      "step": 22496
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2741920183492655e-06,
+      "loss": 0.001,
+      "step": 22497
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2690979261187426e-06,
+      "loss": 0.0003,
+      "step": 22498
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.264006251167513e-06,
+      "loss": 0.0014,
+      "step": 22499
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.258916993580559e-06,
+      "loss": 0.0001,
+      "step": 22500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.253830153442928e-06,
+      "loss": 0.0034,
+      "step": 22501
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2487457308395705e-06,
+      "loss": 0.0009,
+      "step": 22502
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2436637258553685e-06,
+      "loss": 0.0001,
+      "step": 22503
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.238584138575252e-06,
+      "loss": 0.0033,
+      "step": 22504
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.233506969084022e-06,
+      "loss": 0.0001,
+      "step": 22505
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.22843221746651e-06,
+      "loss": 0.0006,
+      "step": 22506
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.223359883807432e-06,
+      "loss": 0.0006,
+      "step": 22507
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.218289968191553e-06,
+      "loss": 0.001,
+      "step": 22508
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.213222470703504e-06,
+      "loss": 0.0006,
+      "step": 22509
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.208157391427953e-06,
+      "loss": 0.0034,
+      "step": 22510
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.203094730449497e-06,
+      "loss": 0.0017,
+      "step": 22511
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.198034487852687e-06,
+      "loss": 0.0009,
+      "step": 22512
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.192976663722004e-06,
+      "loss": 0.0005,
+      "step": 22513
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.187921258141997e-06,
+      "loss": 0.0002,
+      "step": 22514
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.182868271197032e-06,
+      "loss": 0.0085,
+      "step": 22515
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.177817702971526e-06,
+      "loss": 0.0016,
+      "step": 22516
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.172769553549844e-06,
+      "loss": 0.0011,
+      "step": 22517
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.1677238230162675e-06,
+      "loss": 0.0031,
+      "step": 22518
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.162680511455114e-06,
+      "loss": 0.0062,
+      "step": 22519
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.157639618950599e-06,
+      "loss": 0.0038,
+      "step": 22520
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.152601145586904e-06,
+      "loss": 0.0022,
+      "step": 22521
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.147565091448197e-06,
+      "loss": 0.0003,
+      "step": 22522
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.142531456618576e-06,
+      "loss": 0.0001,
+      "step": 22523
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.137500241182108e-06,
+      "loss": 0.004,
+      "step": 22524
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.132471445222841e-06,
+      "loss": 0.0002,
+      "step": 22525
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.127445068824759e-06,
+      "loss": 0.0013,
+      "step": 22526
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.122421112071812e-06,
+      "loss": 0.0023,
+      "step": 22527
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.117399575047898e-06,
+      "loss": 0.0009,
+      "step": 22528
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.112380457836918e-06,
+      "loss": 0.001,
+      "step": 22529
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.10736376052267e-06,
+      "loss": 0.0001,
+      "step": 22530
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.102349483188939e-06,
+      "loss": 0.0039,
+      "step": 22531
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.097337625919507e-06,
+      "loss": 0.0005,
+      "step": 22532
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.0923281887980225e-06,
+      "loss": 0.0034,
+      "step": 22533
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.087321171908237e-06,
+      "loss": 0.0028,
+      "step": 22534
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.082316575333701e-06,
+      "loss": 0.001,
+      "step": 22535
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.077314399158044e-06,
+      "loss": 0.0005,
+      "step": 22536
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.072314643464787e-06,
+      "loss": 0.0003,
+      "step": 22537
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.067317308337443e-06,
+      "loss": 0.0003,
+      "step": 22538
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.06232239385948e-06,
+      "loss": 0.0003,
+      "step": 22539
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.057329900114332e-06,
+      "loss": 0.0007,
+      "step": 22540
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.052339827185364e-06,
+      "loss": 0.0058,
+      "step": 22541
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.047352175155928e-06,
+      "loss": 0.0002,
+      "step": 22542
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.042366944109355e-06,
+      "loss": 0.0031,
+      "step": 22543
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.037384134128847e-06,
+      "loss": 0.0045,
+      "step": 22544
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.032403745297686e-06,
+      "loss": 0.0021,
+      "step": 22545
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.02742577769899e-06,
+      "loss": 0.0011,
+      "step": 22546
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.022450231415992e-06,
+      "loss": 0.0001,
+      "step": 22547
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.017477106531708e-06,
+      "loss": 0.0013,
+      "step": 22548
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.012506403129208e-06,
+      "loss": 0.0016,
+      "step": 22549
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.007538121291587e-06,
+      "loss": 0.005,
+      "step": 22550
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.002572261101717e-06,
+      "loss": 0.0024,
+      "step": 22551
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.997608822642646e-06,
+      "loss": 0.0016,
+      "step": 22552
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.992647805997191e-06,
+      "loss": 0.001,
+      "step": 22553
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.987689211248236e-06,
+      "loss": 0.0004,
+      "step": 22554
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.982733038478598e-06,
+      "loss": 0.0004,
+      "step": 22555
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.977779287771078e-06,
+      "loss": 0.0037,
+      "step": 22556
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.9728279592083756e-06,
+      "loss": 0.0015,
+      "step": 22557
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.967879052873208e-06,
+      "loss": 0.0021,
+      "step": 22558
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.962932568848227e-06,
+      "loss": 0.0001,
+      "step": 22559
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.957988507216048e-06,
+      "loss": 0.0011,
+      "step": 22560
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.953046868059258e-06,
+      "loss": 0.0003,
+      "step": 22561
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.94810765146037e-06,
+      "loss": 0.0021,
+      "step": 22562
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.943170857501904e-06,
+      "loss": 0.0078,
+      "step": 22563
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.9382364862662614e-06,
+      "loss": 0.0011,
+      "step": 22564
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.933304537835942e-06,
+      "loss": 0.0001,
+      "step": 22565
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.928375012293245e-06,
+      "loss": 0.0012,
+      "step": 22566
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.923447909720524e-06,
+      "loss": 0.0008,
+      "step": 22567
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.918523230200078e-06,
+      "loss": 0.0016,
+      "step": 22568
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.913600973814158e-06,
+      "loss": 0.0004,
+      "step": 22569
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.9086811406449655e-06,
+      "loss": 0.001,
+      "step": 22570
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.903763730774684e-06,
+      "loss": 0.002,
+      "step": 22571
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8988487442854325e-06,
+      "loss": 0.0002,
+      "step": 22572
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.893936181259311e-06,
+      "loss": 0.0032,
+      "step": 22573
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.889026041778355e-06,
+      "loss": 0.0005,
+      "step": 22574
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8841183259245795e-06,
+      "loss": 0.0037,
+      "step": 22575
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.879213033779972e-06,
+      "loss": 0.0003,
+      "step": 22576
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.874310165426415e-06,
+      "loss": 0.0019,
+      "step": 22577
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8694097209458275e-06,
+      "loss": 0.0024,
+      "step": 22578
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.864511700420059e-06,
+      "loss": 0.0002,
+      "step": 22579
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8596161039308965e-06,
+      "loss": 0.0002,
+      "step": 22580
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.85472293156014e-06,
+      "loss": 0.0097,
+      "step": 22581
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8498321833894735e-06,
+      "loss": 0.0014,
+      "step": 22582
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.844943859500616e-06,
+      "loss": 0.0002,
+      "step": 22583
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.840057959975169e-06,
+      "loss": 0.0007,
+      "step": 22584
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.835174484894799e-06,
+      "loss": 0.0033,
+      "step": 22585
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.830293434341009e-06,
+      "loss": 0.0008,
+      "step": 22586
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.82541480839535e-06,
+      "loss": 0.0019,
+      "step": 22587
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.82053860713929e-06,
+      "loss": 0.0006,
+      "step": 22588
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.815664830654297e-06,
+      "loss": 0.0012,
+      "step": 22589
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.81079347902174e-06,
+      "loss": 0.001,
+      "step": 22590
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.805924552323004e-06,
+      "loss": 0.0008,
+      "step": 22591
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.801058050639406e-06,
+      "loss": 0.0001,
+      "step": 22592
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.796193974052215e-06,
+      "loss": 0.0012,
+      "step": 22593
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.791332322642699e-06,
+      "loss": 0.0002,
+      "step": 22594
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.786473096491994e-06,
+      "loss": 0.0026,
+      "step": 22595
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.781616295681334e-06,
+      "loss": 0.0032,
+      "step": 22596
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.776761920291772e-06,
+      "loss": 0.0028,
+      "step": 22597
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.771909970404442e-06,
+      "loss": 0.0007,
+      "step": 22598
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.767060446100329e-06,
+      "loss": 0.0009,
+      "step": 22599
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.762213347460469e-06,
+      "loss": 0.0006,
+      "step": 22600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.757368674565798e-06,
+      "loss": 0.0009,
+      "step": 22601
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.752526427497216e-06,
+      "loss": 0.0013,
+      "step": 22602
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.747686606335643e-06,
+      "loss": 0.0026,
+      "step": 22603
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.742849211161881e-06,
+      "loss": 0.003,
+      "step": 22604
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.738014242056715e-06,
+      "loss": 0.0023,
+      "step": 22605
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.7331816991009295e-06,
+      "loss": 0.0023,
+      "step": 22606
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.728351582375211e-06,
+      "loss": 0.0046,
+      "step": 22607
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.723523891960246e-06,
+      "loss": 0.001,
+      "step": 22608
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.718698627936651e-06,
+      "loss": 0.0013,
+      "step": 22609
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.71387579038503e-06,
+      "loss": 0.0001,
+      "step": 22610
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.709055379385934e-06,
+      "loss": 0.0019,
+      "step": 22611
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.704237395019866e-06,
+      "loss": 0.0045,
+      "step": 22612
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.699421837367312e-06,
+      "loss": 0.0042,
+      "step": 22613
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.69460870650869e-06,
+      "loss": 0.0089,
+      "step": 22614
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.689798002524353e-06,
+      "loss": 0.0005,
+      "step": 22615
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.68498972549472e-06,
+      "loss": 0.0019,
+      "step": 22616
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.680183875500043e-06,
+      "loss": 0.0017,
+      "step": 22617
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.675380452620608e-06,
+      "loss": 0.0001,
+      "step": 22618
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.670579456936635e-06,
+      "loss": 0.0002,
+      "step": 22619
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.665780888528308e-06,
+      "loss": 0.0025,
+      "step": 22620
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.660984747475782e-06,
+      "loss": 0.0007,
+      "step": 22621
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.656191033859175e-06,
+      "loss": 0.0008,
+      "step": 22622
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.651399747758505e-06,
+      "loss": 0.0023,
+      "step": 22623
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.646610889253844e-06,
+      "loss": 0.0029,
+      "step": 22624
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.641824458425142e-06,
+      "loss": 0.0055,
+      "step": 22625
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.63704045535237e-06,
+      "loss": 0.0003,
+      "step": 22626
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.632258880115414e-06,
+      "loss": 0.0002,
+      "step": 22627
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.627479732794126e-06,
+      "loss": 0.0008,
+      "step": 22628
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.622703013468343e-06,
+      "loss": 0.0005,
+      "step": 22629
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.617928722217817e-06,
+      "loss": 0.0002,
+      "step": 22630
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.613156859122336e-06,
+      "loss": 0.0005,
+      "step": 22631
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.608387424261567e-06,
+      "loss": 0.0017,
+      "step": 22632
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.6036204177151644e-06,
+      "loss": 0.0003,
+      "step": 22633
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.598855839562782e-06,
+      "loss": 0.0004,
+      "step": 22634
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.594093689883921e-06,
+      "loss": 0.002,
+      "step": 22635
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.589333968758219e-06,
+      "loss": 0.0033,
+      "step": 22636
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.584576676265095e-06,
+      "loss": 0.0027,
+      "step": 22637
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.579821812484053e-06,
+      "loss": 0.0026,
+      "step": 22638
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.575069377494461e-06,
+      "loss": 0.0001,
+      "step": 22639
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.5703193713757245e-06,
+      "loss": 0.0023,
+      "step": 22640
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.565571794207179e-06,
+      "loss": 0.0004,
+      "step": 22641
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.560826646068111e-06,
+      "loss": 0.0002,
+      "step": 22642
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.5560839270377736e-06,
+      "loss": 0.0022,
+      "step": 22643
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.551343637195387e-06,
+      "loss": 0.004,
+      "step": 22644
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.546605776620122e-06,
+      "loss": 0.0013,
+      "step": 22645
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.54187034539108e-06,
+      "loss": 0.0053,
+      "step": 22646
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.537137343587399e-06,
+      "loss": 0.0082,
+      "step": 22647
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.532406771288067e-06,
+      "loss": 0.0006,
+      "step": 22648
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.527678628572168e-06,
+      "loss": 0.0003,
+      "step": 22649
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.522952915518624e-06,
+      "loss": 0.0032,
+      "step": 22650
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.518229632206371e-06,
+      "loss": 0.0002,
+      "step": 22651
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.513508778714297e-06,
+      "loss": 0.0007,
+      "step": 22652
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.508790355121256e-06,
+      "loss": 0.0011,
+      "step": 22653
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.504074361506049e-06,
+      "loss": 0.0028,
+      "step": 22654
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.499360797947432e-06,
+      "loss": 0.003,
+      "step": 22655
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.494649664524158e-06,
+      "loss": 0.0009,
+      "step": 22656
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.489940961314881e-06,
+      "loss": 0.0004,
+      "step": 22657
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.48523468839827e-06,
+      "loss": 0.0012,
+      "step": 22658
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4805308458529144e-06,
+      "loss": 0.0003,
+      "step": 22659
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4758294337574e-06,
+      "loss": 0.0075,
+      "step": 22660
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.471130452190197e-06,
+      "loss": 0.0002,
+      "step": 22661
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.466433901229843e-06,
+      "loss": 0.0003,
+      "step": 22662
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.461739780954743e-06,
+      "loss": 0.0013,
+      "step": 22663
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.457048091443316e-06,
+      "loss": 0.0,
+      "step": 22664
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.452358832773934e-06,
+      "loss": 0.0039,
+      "step": 22665
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.447672005024883e-06,
+      "loss": 0.0024,
+      "step": 22666
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.442987608274468e-06,
+      "loss": 0.0003,
+      "step": 22667
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.438305642600925e-06,
+      "loss": 0.0065,
+      "step": 22668
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.433626108082428e-06,
+      "loss": 0.0008,
+      "step": 22669
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.428949004797161e-06,
+      "loss": 0.0015,
+      "step": 22670
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.424274332823213e-06,
+      "loss": 0.0001,
+      "step": 22671
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.419602092238689e-06,
+      "loss": 0.0,
+      "step": 22672
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.414932283121625e-06,
+      "loss": 0.0001,
+      "step": 22673
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.410264905549993e-06,
+      "loss": 0.0019,
+      "step": 22674
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.405599959601747e-06,
+      "loss": 0.0009,
+      "step": 22675
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.400937445354824e-06,
+      "loss": 0.001,
+      "step": 22676
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.396277362887079e-06,
+      "loss": 0.0029,
+      "step": 22677
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.39161971227635e-06,
+      "loss": 0.0003,
+      "step": 22678
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.386964493600409e-06,
+      "loss": 0.0027,
+      "step": 22679
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3823117069370584e-06,
+      "loss": 0.0006,
+      "step": 22680
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.377661352363954e-06,
+      "loss": 0.0042,
+      "step": 22681
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3730134299588e-06,
+      "loss": 0.0041,
+      "step": 22682
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3683679397992e-06,
+      "loss": 0.0002,
+      "step": 22683
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.363724881962744e-06,
+      "loss": 0.0058,
+      "step": 22684
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.359084256527018e-06,
+      "loss": 0.001,
+      "step": 22685
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.354446063569478e-06,
+      "loss": 0.0009,
+      "step": 22686
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.349810303167628e-06,
+      "loss": 0.0008,
+      "step": 22687
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.345176975398889e-06,
+      "loss": 0.0039,
+      "step": 22688
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.340546080340618e-06,
+      "loss": 0.0007,
+      "step": 22689
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.335917618070184e-06,
+      "loss": 0.0087,
+      "step": 22690
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.331291588664893e-06,
+      "loss": 0.0087,
+      "step": 22691
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.326667992201998e-06,
+      "loss": 0.0085,
+      "step": 22692
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.322046828758724e-06,
+      "loss": 0.0027,
+      "step": 22693
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.31742809841224e-06,
+      "loss": 0.0011,
+      "step": 22694
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.312811801239718e-06,
+      "loss": 0.0014,
+      "step": 22695
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.308197937318264e-06,
+      "loss": 0.0019,
+      "step": 22696
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.303586506724882e-06,
+      "loss": 0.0002,
+      "step": 22697
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.298977509536644e-06,
+      "loss": 0.005,
+      "step": 22698
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2943709458304886e-06,
+      "loss": 0.0015,
+      "step": 22699
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2897668156834205e-06,
+      "loss": 0.0052,
+      "step": 22700
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.285165119172262e-06,
+      "loss": 0.0037,
+      "step": 22701
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.280565856373918e-06,
+      "loss": 0.0017,
+      "step": 22702
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.275969027365178e-06,
+      "loss": 0.0019,
+      "step": 22703
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2713746322228296e-06,
+      "loss": 0.0012,
+      "step": 22704
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.266782671023611e-06,
+      "loss": 0.0003,
+      "step": 22705
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.262193143844211e-06,
+      "loss": 0.0003,
+      "step": 22706
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.257606050761286e-06,
+      "loss": 0.0006,
+      "step": 22707
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.253021391851441e-06,
+      "loss": 0.0137,
+      "step": 22708
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2484391671912644e-06,
+      "loss": 0.0002,
+      "step": 22709
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.243859376857278e-06,
+      "loss": 0.0031,
+      "step": 22710
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.239282020925988e-06,
+      "loss": 0.0015,
+      "step": 22711
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2347070994738e-06,
+      "loss": 0.0041,
+      "step": 22712
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.230134612577185e-06,
+      "loss": 0.0003,
+      "step": 22713
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.225564560312466e-06,
+      "loss": 0.0086,
+      "step": 22714
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.220996942755983e-06,
+      "loss": 0.0002,
+      "step": 22715
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.21643175998404e-06,
+      "loss": 0.003,
+      "step": 22716
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.21186901207286e-06,
+      "loss": 0.0002,
+      "step": 22717
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2073086990986826e-06,
+      "loss": 0.0021,
+      "step": 22718
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.202750821137646e-06,
+      "loss": 0.0013,
+      "step": 22719
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1981953782658725e-06,
+      "loss": 0.0022,
+      "step": 22720
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.193642370559453e-06,
+      "loss": 0.0016,
+      "step": 22721
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.18909179809444e-06,
+      "loss": 0.0031,
+      "step": 22722
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.184543660946843e-06,
+      "loss": 0.0,
+      "step": 22723
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.179997959192599e-06,
+      "loss": 0.0003,
+      "step": 22724
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.175454692907648e-06,
+      "loss": 0.0002,
+      "step": 22725
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.170913862167879e-06,
+      "loss": 0.0004,
+      "step": 22726
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.166375467049099e-06,
+      "loss": 0.0006,
+      "step": 22727
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.161839507627146e-06,
+      "loss": 0.0031,
+      "step": 22728
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.15730598397776e-06,
+      "loss": 0.0003,
+      "step": 22729
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.152774896176647e-06,
+      "loss": 0.0021,
+      "step": 22730
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.14824624429953e-06,
+      "loss": 0.0006,
+      "step": 22731
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.143720028421982e-06,
+      "loss": 0.0013,
+      "step": 22732
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.139196248619658e-06,
+      "loss": 0.0002,
+      "step": 22733
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1346749049680824e-06,
+      "loss": 0.0003,
+      "step": 22734
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.130155997542761e-06,
+      "loss": 0.0027,
+      "step": 22735
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1256395264191995e-06,
+      "loss": 0.0032,
+      "step": 22736
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.121125491672789e-06,
+      "loss": 0.0004,
+      "step": 22737
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.116613893378984e-06,
+      "loss": 0.0029,
+      "step": 22738
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.112104731613092e-06,
+      "loss": 0.0027,
+      "step": 22739
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.107598006450419e-06,
+      "loss": 0.0091,
+      "step": 22740
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.103093717966255e-06,
+      "loss": 0.0003,
+      "step": 22741
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.098591866235856e-06,
+      "loss": 0.0005,
+      "step": 22742
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.09409245133433e-06,
+      "loss": 0.0017,
+      "step": 22743
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.089595473336899e-06,
+      "loss": 0.0001,
+      "step": 22744
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.085100932318653e-06,
+      "loss": 0.0031,
+      "step": 22745
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.080608828354664e-06,
+      "loss": 0.0001,
+      "step": 22746
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.076119161519958e-06,
+      "loss": 0.0001,
+      "step": 22747
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.071631931889474e-06,
+      "loss": 0.0009,
+      "step": 22748
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.067147139538235e-06,
+      "loss": 0.0062,
+      "step": 22749
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0626647845410646e-06,
+      "loss": 0.0008,
+      "step": 22750
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.058184866972919e-06,
+      "loss": 0.0007,
+      "step": 22751
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.053707386908539e-06,
+      "loss": 0.0002,
+      "step": 22752
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.049232344422748e-06,
+      "loss": 0.0032,
+      "step": 22753
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.04475973959027e-06,
+      "loss": 0.0009,
+      "step": 22754
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.040289572485811e-06,
+      "loss": 0.0046,
+      "step": 22755
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.035821843184045e-06,
+      "loss": 0.0002,
+      "step": 22756
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.031356551759579e-06,
+      "loss": 0.0081,
+      "step": 22757
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.026893698287004e-06,
+      "loss": 0.0017,
+      "step": 22758
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.022433282840826e-06,
+      "loss": 0.0002,
+      "step": 22759
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0179753054955854e-06,
+      "loss": 0.006,
+      "step": 22760
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.013519766325707e-06,
+      "loss": 0.0027,
+      "step": 22761
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0090666654056305e-06,
+      "loss": 0.0003,
+      "step": 22762
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0046160028096966e-06,
+      "loss": 0.0022,
+      "step": 22763
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.000167778612279e-06,
+      "loss": 0.0027,
+      "step": 22764
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9957219928876514e-06,
+      "loss": 0.0002,
+      "step": 22765
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.991278645710056e-06,
+      "loss": 0.0032,
+      "step": 22766
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.986837737153731e-06,
+      "loss": 0.0042,
+      "step": 22767
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.982399267292819e-06,
+      "loss": 0.007,
+      "step": 22768
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.977963236201492e-06,
+      "loss": 0.0017,
+      "step": 22769
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.973529643953793e-06,
+      "loss": 0.0002,
+      "step": 22770
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.969098490623812e-06,
+      "loss": 0.001,
+      "step": 22771
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.964669776285523e-06,
+      "loss": 0.0013,
+      "step": 22772
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.960243501012916e-06,
+      "loss": 0.0031,
+      "step": 22773
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.955819664879917e-06,
+      "loss": 0.0001,
+      "step": 22774
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.951398267960399e-06,
+      "loss": 0.0014,
+      "step": 22775
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.946979310328219e-06,
+      "loss": 0.0066,
+      "step": 22776
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.942562792057186e-06,
+      "loss": 0.0019,
+      "step": 22777
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.93814871322104e-06,
+      "loss": 0.0012,
+      "step": 22778
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9337370738935395e-06,
+      "loss": 0.0009,
+      "step": 22779
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.929327874148358e-06,
+      "loss": 0.0011,
+      "step": 22780
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.924921114059088e-06,
+      "loss": 0.0012,
+      "step": 22781
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.920516793699418e-06,
+      "loss": 0.0001,
+      "step": 22782
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.916114913142826e-06,
+      "loss": 0.0101,
+      "step": 22783
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9117154724629005e-06,
+      "loss": 0.0003,
+      "step": 22784
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.907318471733068e-06,
+      "loss": 0.0002,
+      "step": 22785
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.902923911026784e-06,
+      "loss": 0.0012,
+      "step": 22786
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.898531790417459e-06,
+      "loss": 0.0015,
+      "step": 22787
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.894142109978432e-06,
+      "loss": 0.004,
+      "step": 22788
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8897548697830295e-06,
+      "loss": 0.0005,
+      "step": 22789
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.885370069904525e-06,
+      "loss": 0.0004,
+      "step": 22790
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.880987710416145e-06,
+      "loss": 0.0033,
+      "step": 22791
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.876607791391096e-06,
+      "loss": 0.0003,
+      "step": 22792
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8722303129025354e-06,
+      "loss": 0.0003,
+      "step": 22793
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.867855275023523e-06,
+      "loss": 0.0015,
+      "step": 22794
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8634826778272176e-06,
+      "loss": 0.0005,
+      "step": 22795
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.859112521386576e-06,
+      "loss": 0.0003,
+      "step": 22796
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.85474480577464e-06,
+      "loss": 0.0082,
+      "step": 22797
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.850379531064319e-06,
+      "loss": 0.0001,
+      "step": 22798
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.846016697328519e-06,
+      "loss": 0.0002,
+      "step": 22799
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.84165630464015e-06,
+      "loss": 0.002,
+      "step": 22800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.837298353071988e-06,
+      "loss": 0.0026,
+      "step": 22801
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.83294284269689e-06,
+      "loss": 0.0012,
+      "step": 22802
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.828589773587515e-06,
+      "loss": 0.0002,
+      "step": 22803
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.824239145816638e-06,
+      "loss": 0.0009,
+      "step": 22804
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.819890959456867e-06,
+      "loss": 0.001,
+      "step": 22805
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.815545214580862e-06,
+      "loss": 0.0032,
+      "step": 22806
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.811201911261197e-06,
+      "loss": 0.0006,
+      "step": 22807
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8068610495704144e-06,
+      "loss": 0.0035,
+      "step": 22808
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8025226295810064e-06,
+      "loss": 0.0016,
+      "step": 22809
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.798186651365448e-06,
+      "loss": 0.001,
+      "step": 22810
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7938531149961317e-06,
+      "loss": 0.0009,
+      "step": 22811
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7895220205454656e-06,
+      "loss": 0.0048,
+      "step": 22812
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.785193368085776e-06,
+      "loss": 0.0006,
+      "step": 22813
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7808671576893377e-06,
+      "loss": 0.0003,
+      "step": 22814
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7765433894284436e-06,
+      "loss": 0.0041,
+      "step": 22815
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7722220633752853e-06,
+      "loss": 0.0002,
+      "step": 22816
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7679031796020386e-06,
+      "loss": 0.0008,
+      "step": 22817
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.763586738180846e-06,
+      "loss": 0.0021,
+      "step": 22818
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7592727391837994e-06,
+      "loss": 0.0004,
+      "step": 22819
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.754961182682942e-06,
+      "loss": 0.0034,
+      "step": 22820
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7506520687502985e-06,
+      "loss": 0.0021,
+      "step": 22821
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.746345397457812e-06,
+      "loss": 0.0001,
+      "step": 22822
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.742041168877441e-06,
+      "loss": 0.0002,
+      "step": 22823
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.737739383081062e-06,
+      "loss": 0.0002,
+      "step": 22824
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.733440040140534e-06,
+      "loss": 0.0049,
+      "step": 22825
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7291431401276663e-06,
+      "loss": 0.0004,
+      "step": 22826
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7248486831141678e-06,
+      "loss": 0.0001,
+      "step": 22827
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.720556669171848e-06,
+      "loss": 0.0037,
+      "step": 22828
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.716267098372333e-06,
+      "loss": 0.0016,
+      "step": 22829
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7119799707872987e-06,
+      "loss": 0.0029,
+      "step": 22830
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.707695286488355e-06,
+      "loss": 0.0001,
+      "step": 22831
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7034130455470103e-06,
+      "loss": 0.0007,
+      "step": 22832
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6991332480348577e-06,
+      "loss": 0.0037,
+      "step": 22833
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6948558940233077e-06,
+      "loss": 0.0005,
+      "step": 22834
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6905809835838683e-06,
+      "loss": 0.0003,
+      "step": 22835
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6863085167878835e-06,
+      "loss": 0.0012,
+      "step": 22836
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.682038493706746e-06,
+      "loss": 0.001,
+      "step": 22837
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6777709144117486e-06,
+      "loss": 0.0003,
+      "step": 22838
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.673505778974184e-06,
+      "loss": 0.0004,
+      "step": 22839
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.669243087465279e-06,
+      "loss": 0.0001,
+      "step": 22840
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.664982839956243e-06,
+      "loss": 0.0024,
+      "step": 22841
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6607250365182196e-06,
+      "loss": 0.0001,
+      "step": 22842
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.656469677222318e-06,
+      "loss": 0.0019,
+      "step": 22843
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6522167621396147e-06,
+      "loss": 0.0002,
+      "step": 22844
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.647966291341137e-06,
+      "loss": 0.0042,
+      "step": 22845
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.64371826489791e-06,
+      "loss": 0.0002,
+      "step": 22846
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.639472682880812e-06,
+      "loss": 0.0009,
+      "step": 22847
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.635229545360835e-06,
+      "loss": 0.0013,
+      "step": 22848
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6309888524087894e-06,
+      "loss": 0.0037,
+      "step": 22849
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6267506040955183e-06,
+      "loss": 0.0012,
+      "step": 22850
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6225148004918157e-06,
+      "loss": 0.0008,
+      "step": 22851
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.618281441668408e-06,
+      "loss": 0.0001,
+      "step": 22852
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6140505276960384e-06,
+      "loss": 0.0018,
+      "step": 22853
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.609822058645334e-06,
+      "loss": 0.0008,
+      "step": 22854
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6055960345869384e-06,
+      "loss": 0.0005,
+      "step": 22855
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.601372455591428e-06,
+      "loss": 0.0007,
+      "step": 22856
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.597151321729347e-06,
+      "loss": 0.0027,
+      "step": 22857
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.592932633071188e-06,
+      "loss": 0.002,
+      "step": 22858
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5887163896874127e-06,
+      "loss": 0.0004,
+      "step": 22859
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5845025916484473e-06,
+      "loss": 0.0008,
+      "step": 22860
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5802912390246685e-06,
+      "loss": 0.0002,
+      "step": 22861
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5760823318864373e-06,
+      "loss": 0.0078,
+      "step": 22862
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5718758703039806e-06,
+      "loss": 0.0001,
+      "step": 22863
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.567671854347642e-06,
+      "loss": 0.0004,
+      "step": 22864
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5634702840875494e-06,
+      "loss": 0.0038,
+      "step": 22865
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.559271159593963e-06,
+      "loss": 0.006,
+      "step": 22866
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5550744809369593e-06,
+      "loss": 0.0004,
+      "step": 22867
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5508802481866493e-06,
+      "loss": 0.0004,
+      "step": 22868
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5466884614130775e-06,
+      "loss": 0.0019,
+      "step": 22869
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.542499120686254e-06,
+      "loss": 0.0002,
+      "step": 22870
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5383122260761565e-06,
+      "loss": 0.0001,
+      "step": 22871
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.534127777652712e-06,
+      "loss": 0.0003,
+      "step": 22872
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5299457754858153e-06,
+      "loss": 0.0003,
+      "step": 22873
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5257662196452927e-06,
+      "loss": 0.0001,
+      "step": 22874
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.521589110200973e-06,
+      "loss": 0.0021,
+      "step": 22875
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5174144472225994e-06,
+      "loss": 0.0005,
+      "step": 22876
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5132422307799333e-06,
+      "loss": 0.0002,
+      "step": 22877
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.509072460942619e-06,
+      "loss": 0.0002,
+      "step": 22878
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5049051377803338e-06,
+      "loss": 0.0002,
+      "step": 22879
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5007402613626225e-06,
+      "loss": 0.0002,
+      "step": 22880
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.496577831759112e-06,
+      "loss": 0.0003,
+      "step": 22881
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.492417849039314e-06,
+      "loss": 0.0011,
+      "step": 22882
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.488260313272656e-06,
+      "loss": 0.0002,
+      "step": 22883
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4841052245286325e-06,
+      "loss": 0.0013,
+      "step": 22884
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.479952582876605e-06,
+      "loss": 0.0113,
+      "step": 22885
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4758023883859675e-06,
+      "loss": 0.0001,
+      "step": 22886
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.471654641125998e-06,
+      "loss": 0.0021,
+      "step": 22887
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4675093411659915e-06,
+      "loss": 0.0005,
+      "step": 22888
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4633664885751753e-06,
+      "loss": 0.0001,
+      "step": 22889
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.459226083422745e-06,
+      "loss": 0.0019,
+      "step": 22890
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4550881257778606e-06,
+      "loss": 0.0028,
+      "step": 22891
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4509526157096347e-06,
+      "loss": 0.0009,
+      "step": 22892
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4468195532871114e-06,
+      "loss": 0.0015,
+      "step": 22893
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4426889385793523e-06,
+      "loss": 0.0018,
+      "step": 22894
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.438560771655352e-06,
+      "loss": 0.0031,
+      "step": 22895
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4344350525840057e-06,
+      "loss": 0.0006,
+      "step": 22896
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.430311781434292e-06,
+      "loss": 0.0002,
+      "step": 22897
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.426190958275005e-06,
+      "loss": 0.0005,
+      "step": 22898
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4220725831750393e-06,
+      "loss": 0.0001,
+      "step": 22899
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4179566562031245e-06,
+      "loss": 0.0002,
+      "step": 22900
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4138431774280384e-06,
+      "loss": 0.0009,
+      "step": 22901
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.409732146918459e-06,
+      "loss": 0.0002,
+      "step": 22902
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.405623564743065e-06,
+      "loss": 0.0002,
+      "step": 22903
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4015174309704853e-06,
+      "loss": 0.0036,
+      "step": 22904
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3974137456692806e-06,
+      "loss": 0.0006,
+      "step": 22905
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.393312508907997e-06,
+      "loss": 0.0008,
+      "step": 22906
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3892137207551463e-06,
+      "loss": 0.0017,
+      "step": 22907
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3851173812791565e-06,
+      "loss": 0.0035,
+      "step": 22908
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3810234905484735e-06,
+      "loss": 0.0002,
+      "step": 22909
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3769320486314754e-06,
+      "loss": 0.0007,
+      "step": 22910
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3728430555964414e-06,
+      "loss": 0.0035,
+      "step": 22911
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3687565115117334e-06,
+      "loss": 0.0025,
+      "step": 22912
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3646724164455796e-06,
+      "loss": 0.0001,
+      "step": 22913
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3605907704661763e-06,
+      "loss": 0.0028,
+      "step": 22914
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3565115736417345e-06,
+      "loss": 0.0002,
+      "step": 22915
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3524348260403177e-06,
+      "loss": 0.0003,
+      "step": 22916
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.348360527730104e-06,
+      "loss": 0.0003,
+      "step": 22917
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3442886787790722e-06,
+      "loss": 0.0021,
+      "step": 22918
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3402192792552516e-06,
+      "loss": 0.0045,
+      "step": 22919
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3361523292266047e-06,
+      "loss": 0.0012,
+      "step": 22920
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3320878287610597e-06,
+      "loss": 0.0004,
+      "step": 22921
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.328025777926513e-06,
+      "loss": 0.0018,
+      "step": 22922
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3239661767907764e-06,
+      "loss": 0.0003,
+      "step": 22923
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3199090254216964e-06,
+      "loss": 0.0003,
+      "step": 22924
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3158543238870183e-06,
+      "loss": 0.0001,
+      "step": 22925
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.311802072254438e-06,
+      "loss": 0.0009,
+      "step": 22926
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.307752270591668e-06,
+      "loss": 0.0001,
+      "step": 22927
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3037049189663546e-06,
+      "loss": 0.0012,
+      "step": 22928
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.299660017446043e-06,
+      "loss": 0.0031,
+      "step": 22929
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2956175660983632e-06,
+      "loss": 0.0001,
+      "step": 22930
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2915775649907604e-06,
+      "loss": 0.0048,
+      "step": 22931
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.287540014190765e-06,
+      "loss": 0.0042,
+      "step": 22932
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.283504913765789e-06,
+      "loss": 0.0043,
+      "step": 22933
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2794722637832115e-06,
+      "loss": 0.0027,
+      "step": 22934
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2754420643104297e-06,
+      "loss": 0.0002,
+      "step": 22935
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2714143154147053e-06,
+      "loss": 0.0006,
+      "step": 22936
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2673890171633356e-06,
+      "loss": 0.0048,
+      "step": 22937
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.263366169623549e-06,
+      "loss": 0.0028,
+      "step": 22938
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.259345772862526e-06,
+      "loss": 0.0001,
+      "step": 22939
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.255327826947429e-06,
+      "loss": 0.0008,
+      "step": 22940
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.251312331945355e-06,
+      "loss": 0.009,
+      "step": 22941
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2472992879233827e-06,
+      "loss": 0.0026,
+      "step": 22942
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2432886949485094e-06,
+      "loss": 0.0015,
+      "step": 22943
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2392805530877477e-06,
+      "loss": 0.001,
+      "step": 22944
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.235274862408027e-06,
+      "loss": 0.0001,
+      "step": 22945
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.231271622976278e-06,
+      "loss": 0.0042,
+      "step": 22946
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.227270834859297e-06,
+      "loss": 0.0009,
+      "step": 22947
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.22327249812398e-06,
+      "loss": 0.0001,
+      "step": 22948
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.219276612837057e-06,
+      "loss": 0.0005,
+      "step": 22949
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2152831790652923e-06,
+      "loss": 0.0001,
+      "step": 22950
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2112921968753646e-06,
+      "loss": 0.0012,
+      "step": 22951
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.207303666333938e-06,
+      "loss": 0.0001,
+      "step": 22952
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2033175875076424e-06,
+      "loss": 0.0003,
+      "step": 22953
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1993339604630407e-06,
+      "loss": 0.0014,
+      "step": 22954
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.195352785266664e-06,
+      "loss": 0.003,
+      "step": 22955
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.191374061985008e-06,
+      "loss": 0.0068,
+      "step": 22956
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1873977906845204e-06,
+      "loss": 0.0024,
+      "step": 22957
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.183423971431631e-06,
+      "loss": 0.0003,
+      "step": 22958
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.179452604292687e-06,
+      "loss": 0.0007,
+      "step": 22959
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1754836893340185e-06,
+      "loss": 0.0047,
+      "step": 22960
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1715172266219556e-06,
+      "loss": 0.004,
+      "step": 22961
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.167553216222679e-06,
+      "loss": 0.0001,
+      "step": 22962
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1635916582024524e-06,
+      "loss": 0.0014,
+      "step": 22963
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1596325526274224e-06,
+      "loss": 0.0021,
+      "step": 22964
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.15567589956367e-06,
+      "loss": 0.0027,
+      "step": 22965
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1517216990773585e-06,
+      "loss": 0.0027,
+      "step": 22966
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.147769951234469e-06,
+      "loss": 0.0015,
+      "step": 22967
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1438206561010483e-06,
+      "loss": 0.0005,
+      "step": 22968
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.139873813743027e-06,
+      "loss": 0.0003,
+      "step": 22969
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.135929424226319e-06,
+      "loss": 0.0035,
+      "step": 22970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1319874876168215e-06,
+      "loss": 0.0013,
+      "step": 22971
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.128048003980349e-06,
+      "loss": 0.0002,
+      "step": 22972
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.124110973382732e-06,
+      "loss": 0.0002,
+      "step": 22973
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1201763958897173e-06,
+      "loss": 0.0003,
+      "step": 22974
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1162442715669866e-06,
+      "loss": 0.001,
+      "step": 22975
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1123146004802536e-06,
+      "loss": 0.0001,
+      "step": 22976
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.108387382695149e-06,
+      "loss": 0.0022,
+      "step": 22977
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.104462618277237e-06,
+      "loss": 0.0029,
+      "step": 22978
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.100540307292099e-06,
+      "loss": 0.0001,
+      "step": 22979
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.096620449805215e-06,
+      "loss": 0.0007,
+      "step": 22980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0927030458820836e-06,
+      "loss": 0.001,
+      "step": 22981
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.088788095588102e-06,
+      "loss": 0.0014,
+      "step": 22982
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.084875598988684e-06,
+      "loss": 0.0019,
+      "step": 22983
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.080965556149162e-06,
+      "loss": 0.0006,
+      "step": 22984
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0770579671348496e-06,
+      "loss": 0.0002,
+      "step": 22985
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0731528320109946e-06,
+      "loss": 0.0009,
+      "step": 22986
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0692501508428446e-06,
+      "loss": 0.0017,
+      "step": 22987
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.065349923695565e-06,
+      "loss": 0.0002,
+      "step": 22988
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.061452150634286e-06,
+      "loss": 0.0054,
+      "step": 22989
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0575568317241394e-06,
+      "loss": 0.0051,
+      "step": 22990
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0536639670301564e-06,
+      "loss": 0.0006,
+      "step": 22991
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.049773556617385e-06,
+      "loss": 0.0048,
+      "step": 22992
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.045885600550757e-06,
+      "loss": 0.0,
+      "step": 22993
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0420000988952532e-06,
+      "loss": 0.0019,
+      "step": 22994
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.038117051715738e-06,
+      "loss": 0.0052,
+      "step": 22995
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0342364590770774e-06,
+      "loss": 0.0004,
+      "step": 22996
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0303583210441183e-06,
+      "loss": 0.0036,
+      "step": 22997
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0264826376815597e-06,
+      "loss": 0.0022,
+      "step": 22998
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0226094090541997e-06,
+      "loss": 0.0005,
+      "step": 22999
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0187386352266695e-06,
+      "loss": 0.0042,
+      "step": 23000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0148703162636844e-06,
+      "loss": 0.0001,
+      "step": 23001
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0110044522298095e-06,
+      "loss": 0.0041,
+      "step": 23002
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.007141043189626e-06,
+      "loss": 0.0028,
+      "step": 23003
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.003280089207649e-06,
+      "loss": 0.0017,
+      "step": 23004
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.999421590348361e-06,
+      "loss": 0.0013,
+      "step": 23005
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.995565546676226e-06,
+      "loss": 0.0003,
+      "step": 23006
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9917119582556435e-06,
+      "loss": 0.0007,
+      "step": 23007
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9878608251509615e-06,
+      "loss": 0.0006,
+      "step": 23008
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9840121474265123e-06,
+      "loss": 0.0001,
+      "step": 23009
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.980165925146577e-06,
+      "loss": 0.0006,
+      "step": 23010
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9763221583753714e-06,
+      "loss": 0.004,
+      "step": 23011
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9724808471771277e-06,
+      "loss": 0.0001,
+      "step": 23012
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9686419916159774e-06,
+      "loss": 0.0005,
+      "step": 23013
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9648055917560697e-06,
+      "loss": 0.0001,
+      "step": 23014
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9609716476614365e-06,
+      "loss": 0.0015,
+      "step": 23015
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.957140159396143e-06,
+      "loss": 0.0043,
+      "step": 23016
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.953311127024155e-06,
+      "loss": 0.0006,
+      "step": 23017
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9494845506094376e-06,
+      "loss": 0.0001,
+      "step": 23018
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9456604302159403e-06,
+      "loss": 0.0029,
+      "step": 23019
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9418387659074783e-06,
+      "loss": 0.0024,
+      "step": 23020
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9380195577479006e-06,
+      "loss": 0.0009,
+      "step": 23021
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.934202805800989e-06,
+      "loss": 0.0007,
+      "step": 23022
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9303885101305104e-06,
+      "loss": 0.0001,
+      "step": 23023
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.926576670800146e-06,
+      "loss": 0.0001,
+      "step": 23024
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9227672878735788e-06,
+      "loss": 0.0002,
+      "step": 23025
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.918960361414424e-06,
+      "loss": 0.0006,
+      "step": 23026
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.915155891486265e-06,
+      "loss": 0.0013,
+      "step": 23027
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.91135387815265e-06,
+      "loss": 0.0007,
+      "step": 23028
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.907554321477079e-06,
+      "loss": 0.0028,
+      "step": 23029
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.903757221523001e-06,
+      "loss": 0.0013,
+      "step": 23030
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8999625783538317e-06,
+      "loss": 0.0105,
+      "step": 23031
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8961703920329704e-06,
+      "loss": 0.0031,
+      "step": 23032
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8923806626237334e-06,
+      "loss": 0.002,
+      "step": 23033
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8885933901894365e-06,
+      "loss": 0.0029,
+      "step": 23034
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8848085747933127e-06,
+      "loss": 0.0065,
+      "step": 23035
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8810262164985943e-06,
+      "loss": 0.0003,
+      "step": 23036
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.877246315368431e-06,
+      "loss": 0.0011,
+      "step": 23037
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.873468871465989e-06,
+      "loss": 0.0005,
+      "step": 23038
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8696938848543184e-06,
+      "loss": 0.0004,
+      "step": 23039
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8659213555964844e-06,
+      "loss": 0.0008,
+      "step": 23040
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8621512837555207e-06,
+      "loss": 0.0063,
+      "step": 23041
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8583836693943595e-06,
+      "loss": 0.0,
+      "step": 23042
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8546185125759515e-06,
+      "loss": 0.0004,
+      "step": 23043
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8508558133631453e-06,
+      "loss": 0.0025,
+      "step": 23044
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8470955718188415e-06,
+      "loss": 0.0013,
+      "step": 23045
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8433377880057893e-06,
+      "loss": 0.0036,
+      "step": 23046
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8395824619867723e-06,
+      "loss": 0.0001,
+      "step": 23047
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.83582959382454e-06,
+      "loss": 0.0045,
+      "step": 23048
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.832079183581709e-06,
+      "loss": 0.0001,
+      "step": 23049
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.828331231320996e-06,
+      "loss": 0.0001,
+      "step": 23050
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8245857371049176e-06,
+      "loss": 0.0022,
+      "step": 23051
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8208427009961066e-06,
+      "loss": 0.001,
+      "step": 23052
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8171021230570136e-06,
+      "loss": 0.0034,
+      "step": 23053
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.813364003350155e-06,
+      "loss": 0.0003,
+      "step": 23054
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8096283419379475e-06,
+      "loss": 0.004,
+      "step": 23055
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.805895138882791e-06,
+      "loss": 0.0002,
+      "step": 23056
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.802164394247036e-06,
+      "loss": 0.0005,
+      "step": 23057
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7984361080929994e-06,
+      "loss": 0.0026,
+      "step": 23058
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7947102804829313e-06,
+      "loss": 0.0081,
+      "step": 23059
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.790986911479065e-06,
+      "loss": 0.0017,
+      "step": 23060
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7872660011436176e-06,
+      "loss": 0.002,
+      "step": 23061
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7835475495386894e-06,
+      "loss": 0.0004,
+      "step": 23062
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7798315567264304e-06,
+      "loss": 0.0022,
+      "step": 23063
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7761180227688416e-06,
+      "loss": 0.0026,
+      "step": 23064
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.772406947728023e-06,
+      "loss": 0.0014,
+      "step": 23065
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7686983316659083e-06,
+      "loss": 0.0004,
+      "step": 23066
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7649921746444646e-06,
+      "loss": 0.0003,
+      "step": 23067
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7612884767255594e-06,
+      "loss": 0.0022,
+      "step": 23068
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.757587237971076e-06,
+      "loss": 0.0001,
+      "step": 23069
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7538884584428156e-06,
+      "loss": 0.0007,
+      "step": 23070
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.750192138202578e-06,
+      "loss": 0.0003,
+      "step": 23071
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7464982773120646e-06,
+      "loss": 0.0001,
+      "step": 23072
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.742806875833009e-06,
+      "loss": 0.0013,
+      "step": 23073
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7391179338270454e-06,
+      "loss": 0.0017,
+      "step": 23074
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7354314513557906e-06,
+      "loss": 0.0035,
+      "step": 23075
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.731747428480813e-06,
+      "loss": 0.0004,
+      "step": 23076
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7280658652636122e-06,
+      "loss": 0.0001,
+      "step": 23077
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.72438676176574e-06,
+      "loss": 0.0006,
+      "step": 23078
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.720710118048597e-06,
+      "loss": 0.0019,
+      "step": 23079
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7170359341736004e-06,
+      "loss": 0.0001,
+      "step": 23080
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7133642102021513e-06,
+      "loss": 0.0001,
+      "step": 23081
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.709694946195501e-06,
+      "loss": 0.0007,
+      "step": 23082
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.706028142215e-06,
+      "loss": 0.0003,
+      "step": 23083
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.702363798321866e-06,
+      "loss": 0.0002,
+      "step": 23084
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6987019145773004e-06,
+      "loss": 0.007,
+      "step": 23085
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.695042491042454e-06,
+      "loss": 0.0004,
+      "step": 23086
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6913855277784613e-06,
+      "loss": 0.002,
+      "step": 23087
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.68773102484639e-06,
+      "loss": 0.0018,
+      "step": 23088
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6840789823072752e-06,
+      "loss": 0.001,
+      "step": 23089
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6804294002221337e-06,
+      "loss": 0.0027,
+      "step": 23090
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.676782278651901e-06,
+      "loss": 0.0021,
+      "step": 23091
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6731376176574948e-06,
+      "loss": 0.0002,
+      "step": 23092
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6694954172997995e-06,
+      "loss": 0.0004,
+      "step": 23093
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6658556776396333e-06,
+      "loss": 0.0001,
+      "step": 23094
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.662218398737781e-06,
+      "loss": 0.002,
+      "step": 23095
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.658583580655027e-06,
+      "loss": 0.0001,
+      "step": 23096
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.654951223452023e-06,
+      "loss": 0.0019,
+      "step": 23097
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6513213271894875e-06,
+      "loss": 0.0003,
+      "step": 23098
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6476938919280043e-06,
+      "loss": 0.0006,
+      "step": 23099
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6440689177281926e-06,
+      "loss": 0.0001,
+      "step": 23100
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.640446404650587e-06,
+      "loss": 0.0001,
+      "step": 23101
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.636826352755672e-06,
+      "loss": 0.0003,
+      "step": 23102
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.63320876210395e-06,
+      "loss": 0.0012,
+      "step": 23103
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6295936327558054e-06,
+      "loss": 0.0008,
+      "step": 23104
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6259809647716236e-06,
+      "loss": 0.0032,
+      "step": 23105
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.622370758211756e-06,
+      "loss": 0.0002,
+      "step": 23106
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6187630131364877e-06,
+      "loss": 0.0003,
+      "step": 23107
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.615157729606071e-06,
+      "loss": 0.0004,
+      "step": 23108
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6115549076807406e-06,
+      "loss": 0.0005,
+      "step": 23109
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6079545474206487e-06,
+      "loss": 0.005,
+      "step": 23110
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6043566488859468e-06,
+      "loss": 0.0037,
+      "step": 23111
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6007612121367205e-06,
+      "loss": 0.0023,
+      "step": 23112
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.597168237233005e-06,
+      "loss": 0.0003,
+      "step": 23113
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5935777242348353e-06,
+      "loss": 0.004,
+      "step": 23114
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5899896732021473e-06,
+      "loss": 0.0021,
+      "step": 23115
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.58640408419491e-06,
+      "loss": 0.0012,
+      "step": 23116
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.582820957272991e-06,
+      "loss": 0.001,
+      "step": 23117
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5792402924962095e-06,
+      "loss": 0.0011,
+      "step": 23118
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.575662089924402e-06,
+      "loss": 0.0046,
+      "step": 23119
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5720863496173196e-06,
+      "loss": 0.006,
+      "step": 23120
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5685130716346815e-06,
+      "loss": 0.0006,
+      "step": 23121
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.564942256036173e-06,
+      "loss": 0.0005,
+      "step": 23122
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.561373902881447e-06,
+      "loss": 0.0002,
+      "step": 23123
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5578080122300725e-06,
+      "loss": 0.0009,
+      "step": 23124
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.554244584141618e-06,
+      "loss": 0.0033,
+      "step": 23125
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.550683618675603e-06,
+      "loss": 0.0013,
+      "step": 23126
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5471251158915296e-06,
+      "loss": 0.0001,
+      "step": 23127
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.543569075848767e-06,
+      "loss": 0.0002,
+      "step": 23128
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5400154986067677e-06,
+      "loss": 0.0002,
+      "step": 23129
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5364643842248345e-06,
+      "loss": 0.0052,
+      "step": 23130
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5329157327623197e-06,
+      "loss": 0.0008,
+      "step": 23131
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.529369544278492e-06,
+      "loss": 0.0003,
+      "step": 23132
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.525825818832539e-06,
+      "loss": 0.0039,
+      "step": 23133
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.522284556483678e-06,
+      "loss": 0.0041,
+      "step": 23134
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.518745757291063e-06,
+      "loss": 0.0009,
+      "step": 23135
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.515209421313763e-06,
+      "loss": 0.0001,
+      "step": 23136
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.511675548610881e-06,
+      "loss": 0.0215,
+      "step": 23137
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5081441392414026e-06,
+      "loss": 0.006,
+      "step": 23138
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5046151932643476e-06,
+      "loss": 0.0005,
+      "step": 23139
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.501088710738619e-06,
+      "loss": 0.0008,
+      "step": 23140
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.497564691723136e-06,
+      "loss": 0.0014,
+      "step": 23141
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.494043136276769e-06,
+      "loss": 0.0017,
+      "step": 23142
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.490524044458303e-06,
+      "loss": 0.0056,
+      "step": 23143
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.487007416326525e-06,
+      "loss": 0.0002,
+      "step": 23144
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.483493251940205e-06,
+      "loss": 0.0002,
+      "step": 23145
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4799815513579625e-06,
+      "loss": 0.0029,
+      "step": 23146
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.476472314638517e-06,
+      "loss": 0.0001,
+      "step": 23147
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.472965541840438e-06,
+      "loss": 0.001,
+      "step": 23148
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.469461233022346e-06,
+      "loss": 0.0031,
+      "step": 23149
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4659593882427108e-06,
+      "loss": 0.0001,
+      "step": 23150
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4624600075600353e-06,
+      "loss": 0.0011,
+      "step": 23151
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4589630910327896e-06,
+      "loss": 0.0042,
+      "step": 23152
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4554686387193436e-06,
+      "loss": 0.0064,
+      "step": 23153
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4519766506781e-06,
+      "loss": 0.0019,
+      "step": 23154
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4484871269673633e-06,
+      "loss": 0.0037,
+      "step": 23155
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.445000067645403e-06,
+      "loss": 0.0002,
+      "step": 23156
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4415154727704888e-06,
+      "loss": 0.0076,
+      "step": 23157
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4380333424007912e-06,
+      "loss": 0.0006,
+      "step": 23158
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.434553676594481e-06,
+      "loss": 0.0,
+      "step": 23159
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.43107647540966e-06,
+      "loss": 0.0005,
+      "step": 23160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4276017389044334e-06,
+      "loss": 0.0034,
+      "step": 23161
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4241294671368205e-06,
+      "loss": 0.0004,
+      "step": 23162
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4206596601648256e-06,
+      "loss": 0.0043,
+      "step": 23163
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.417192318046368e-06,
+      "loss": 0.0004,
+      "step": 23164
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4137274408394023e-06,
+      "loss": 0.0015,
+      "step": 23165
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.410265028601749e-06,
+      "loss": 0.0008,
+      "step": 23166
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.406805081391294e-06,
+      "loss": 0.0007,
+      "step": 23167
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.403347599265776e-06,
+      "loss": 0.0007,
+      "step": 23168
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3998925822829807e-06,
+      "loss": 0.0002,
+      "step": 23169
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3964400305005793e-06,
+      "loss": 0.0014,
+      "step": 23170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.392989943976242e-06,
+      "loss": 0.0001,
+      "step": 23171
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3895423227676238e-06,
+      "loss": 0.0015,
+      "step": 23172
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3860971669322606e-06,
+      "loss": 0.002,
+      "step": 23173
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.38265447652774e-06,
+      "loss": 0.0009,
+      "step": 23174
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3792142516115164e-06,
+      "loss": 0.0055,
+      "step": 23175
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.375776492241077e-06,
+      "loss": 0.0,
+      "step": 23176
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3723411984738263e-06,
+      "loss": 0.0021,
+      "step": 23177
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3689083703671676e-06,
+      "loss": 0.0049,
+      "step": 23178
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3654780079783718e-06,
+      "loss": 0.0052,
+      "step": 23179
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3620501113648105e-06,
+      "loss": 0.0012,
+      "step": 23180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3586246805836704e-06,
+      "loss": 0.0029,
+      "step": 23181
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3552017156922066e-06,
+      "loss": 0.0045,
+      "step": 23182
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3517812167475724e-06,
+      "loss": 0.0058,
+      "step": 23183
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3483631838068893e-06,
+      "loss": 0.0028,
+      "step": 23184
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3449476169272784e-06,
+      "loss": 0.0021,
+      "step": 23185
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3415345161657273e-06,
+      "loss": 0.0036,
+      "step": 23186
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3381238815792903e-06,
+      "loss": 0.0027,
+      "step": 23187
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3347157132249215e-06,
+      "loss": 0.0017,
+      "step": 23188
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3313100111595264e-06,
+      "loss": 0.0001,
+      "step": 23189
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.327906775440008e-06,
+      "loss": 0.0002,
+      "step": 23190
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.324506006123189e-06,
+      "loss": 0.0009,
+      "step": 23191
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3211077032658898e-06,
+      "loss": 0.0016,
+      "step": 23192
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3177118669248485e-06,
+      "loss": 0.0008,
+      "step": 23193
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3143184971567863e-06,
+      "loss": 0.0056,
+      "step": 23194
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.310927594018375e-06,
+      "loss": 0.0029,
+      "step": 23195
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.307539157566285e-06,
+      "loss": 0.0001,
+      "step": 23196
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3041531878570385e-06,
+      "loss": 0.0011,
+      "step": 23197
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3007696849472566e-06,
+      "loss": 0.0001,
+      "step": 23198
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.297388648893411e-06,
+      "loss": 0.0071,
+      "step": 23199
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2940100797520067e-06,
+      "loss": 0.0003,
+      "step": 23200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.290633977579431e-06,
+      "loss": 0.0019,
+      "step": 23201
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.28726034243209e-06,
+      "loss": 0.002,
+      "step": 23202
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.283889174366321e-06,
+      "loss": 0.0001,
+      "step": 23203
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2805204734384463e-06,
+      "loss": 0.0002,
+      "step": 23204
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2771542397047205e-06,
+      "loss": 0.0048,
+      "step": 23205
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2737904732213486e-06,
+      "loss": 0.0042,
+      "step": 23206
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2704291740445523e-06,
+      "loss": 0.0016,
+      "step": 23207
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.267070342230437e-06,
+      "loss": 0.0019,
+      "step": 23208
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2637139778351243e-06,
+      "loss": 0.0023,
+      "step": 23209
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.260360080914636e-06,
+      "loss": 0.0004,
+      "step": 23210
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2570086515250443e-06,
+      "loss": 0.0019,
+      "step": 23211
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2536596897222535e-06,
+      "loss": 0.0018,
+      "step": 23212
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2503131955622698e-06,
+      "loss": 0.0003,
+      "step": 23213
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2469691691009482e-06,
+      "loss": 0.0004,
+      "step": 23214
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2436276103941274e-06,
+      "loss": 0.0004,
+      "step": 23215
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.240288519497646e-06,
+      "loss": 0.0009,
+      "step": 23216
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2369518964672593e-06,
+      "loss": 0.0068,
+      "step": 23217
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2336177413587066e-06,
+      "loss": 0.0015,
+      "step": 23218
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2302860542276758e-06,
+      "loss": 0.0009,
+      "step": 23219
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2269568351297894e-06,
+      "loss": 0.0095,
+      "step": 23220
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2236300841206535e-06,
+      "loss": 0.0058,
+      "step": 23221
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.220305801255856e-06,
+      "loss": 0.0005,
+      "step": 23222
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2169839865909036e-06,
+      "loss": 0.0007,
+      "step": 23223
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2136646401812842e-06,
+      "loss": 0.0057,
+      "step": 23224
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2103477620824206e-06,
+      "loss": 0.0001,
+      "step": 23225
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2070333523497353e-06,
+      "loss": 0.0004,
+      "step": 23226
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.203721411038567e-06,
+      "loss": 0.0004,
+      "step": 23227
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.200411938204222e-06,
+      "loss": 0.0014,
+      "step": 23228
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.197104933902022e-06,
+      "loss": 0.0008,
+      "step": 23229
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.193800398187123e-06,
+      "loss": 0.0016,
+      "step": 23230
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1904983311147984e-06,
+      "loss": 0.0006,
+      "step": 23231
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.187198732740153e-06,
+      "loss": 0.0005,
+      "step": 23232
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1839016031183097e-06,
+      "loss": 0.0103,
+      "step": 23233
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1806069423043084e-06,
+      "loss": 0.0002,
+      "step": 23234
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1773147503532206e-06,
+      "loss": 0.0083,
+      "step": 23235
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1740250273200032e-06,
+      "loss": 0.0002,
+      "step": 23236
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.170737773259612e-06,
+      "loss": 0.0015,
+      "step": 23237
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1674529882269533e-06,
+      "loss": 0.0007,
+      "step": 23238
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1641706722768827e-06,
+      "loss": 0.0042,
+      "step": 23239
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1608908254642067e-06,
+      "loss": 0.0003,
+      "step": 23240
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1576134478437313e-06,
+      "loss": 0.0004,
+      "step": 23241
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.154338539470196e-06,
+      "loss": 0.0007,
+      "step": 23242
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.151066100398258e-06,
+      "loss": 0.0014,
+      "step": 23243
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.147796130682622e-06,
+      "loss": 0.0007,
+      "step": 23244
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.144528630377862e-06,
+      "loss": 0.0002,
+      "step": 23245
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.141263599538584e-06,
+      "loss": 0.0004,
+      "step": 23246
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1380010382193115e-06,
+      "loss": 0.0014,
+      "step": 23247
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.134740946474517e-06,
+      "loss": 0.0097,
+      "step": 23248
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1314833243586904e-06,
+      "loss": 0.0001,
+      "step": 23249
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1282281719261886e-06,
+      "loss": 0.0043,
+      "step": 23250
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1249754892314174e-06,
+      "loss": 0.0004,
+      "step": 23251
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.121725276328684e-06,
+      "loss": 0.0002,
+      "step": 23252
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.118477533272278e-06,
+      "loss": 0.0001,
+      "step": 23253
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1152322601164397e-06,
+      "loss": 0.0003,
+      "step": 23254
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.111989456915375e-06,
+      "loss": 0.0002,
+      "step": 23255
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.108749123723258e-06,
+      "loss": 0.0004,
+      "step": 23256
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.105511260594178e-06,
+      "loss": 0.001,
+      "step": 23257
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.102275867582243e-06,
+      "loss": 0.0008,
+      "step": 23258
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.099042944741475e-06,
+      "loss": 0.0005,
+      "step": 23259
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0958124921258813e-06,
+      "loss": 0.0001,
+      "step": 23260
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0925845097893857e-06,
+      "loss": 0.0007,
+      "step": 23261
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.089358997785945e-06,
+      "loss": 0.0018,
+      "step": 23262
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0861359561693822e-06,
+      "loss": 0.0,
+      "step": 23263
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0829153849935886e-06,
+      "loss": 0.005,
+      "step": 23264
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0796972843123037e-06,
+      "loss": 0.0011,
+      "step": 23265
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.076481654179285e-06,
+      "loss": 0.001,
+      "step": 23266
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0732684946482724e-06,
+      "loss": 0.0037,
+      "step": 23267
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.07005780577289e-06,
+      "loss": 0.0012,
+      "step": 23268
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0668495876067948e-06,
+      "loss": 0.0001,
+      "step": 23269
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.063643840203544e-06,
+      "loss": 0.0011,
+      "step": 23270
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0604405636166943e-06,
+      "loss": 0.0014,
+      "step": 23271
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0572397578997367e-06,
+      "loss": 0.0003,
+      "step": 23272
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0540414231061287e-06,
+      "loss": 0.0012,
+      "step": 23273
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.05084555928931e-06,
+      "loss": 0.0004,
+      "step": 23274
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.047652166502656e-06,
+      "loss": 0.0093,
+      "step": 23275
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0444612447994726e-06,
+      "loss": 0.0001,
+      "step": 23276
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0412727942330675e-06,
+      "loss": 0.0013,
+      "step": 23277
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0380868148567152e-06,
+      "loss": 0.0011,
+      "step": 23278
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0349033067236064e-06,
+      "loss": 0.0008,
+      "step": 23279
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0317222698869152e-06,
+      "loss": 0.0011,
+      "step": 23280
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0285437043997655e-06,
+      "loss": 0.0025,
+      "step": 23281
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.025367610315265e-06,
+      "loss": 0.0008,
+      "step": 23282
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.022193987686438e-06,
+      "loss": 0.0002,
+      "step": 23283
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0190228365663088e-06,
+      "loss": 0.0032,
+      "step": 23284
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0158541570078178e-06,
+      "loss": 0.0001,
+      "step": 23285
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.012687949063907e-06,
+      "loss": 0.0002,
+      "step": 23286
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.00952421278745e-06,
+      "loss": 0.0059,
+      "step": 23287
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.006362948231305e-06,
+      "loss": 0.002,
+      "step": 23288
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0032041554482457e-06,
+      "loss": 0.0017,
+      "step": 23289
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0000478344910474e-06,
+      "loss": 0.0011,
+      "step": 23290
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9968939854124013e-06,
+      "loss": 0.0003,
+      "step": 23291
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.993742608265014e-06,
+      "loss": 0.0001,
+      "step": 23292
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9905937031015285e-06,
+      "loss": 0.0016,
+      "step": 23293
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.987447269974485e-06,
+      "loss": 0.0001,
+      "step": 23294
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9843033089364745e-06,
+      "loss": 0.0004,
+      "step": 23295
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9811618200399895e-06,
+      "loss": 0.0007,
+      "step": 23296
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.97802280333752e-06,
+      "loss": 0.0001,
+      "step": 23297
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9748862588814918e-06,
+      "loss": 0.0002,
+      "step": 23298
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9717521867242457e-06,
+      "loss": 0.005,
+      "step": 23299
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9686205869181905e-06,
+      "loss": 0.0003,
+      "step": 23300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9654914595155836e-06,
+      "loss": 0.0031,
+      "step": 23301
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9623648045687334e-06,
+      "loss": 0.0068,
+      "step": 23302
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9592406221297985e-06,
+      "loss": 0.0011,
+      "step": 23303
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9561189122510036e-06,
+      "loss": 0.0003,
+      "step": 23304
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9529996749844734e-06,
+      "loss": 0.0031,
+      "step": 23305
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9498829103823166e-06,
+      "loss": 0.0021,
+      "step": 23306
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9467686184965748e-06,
+      "loss": 0.0022,
+      "step": 23307
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9436567993792563e-06,
+      "loss": 0.0018,
+      "step": 23308
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9405474530823527e-06,
+      "loss": 0.014,
+      "step": 23309
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9374405796577895e-06,
+      "loss": 0.0001,
+      "step": 23310
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.934336179157475e-06,
+      "loss": 0.0031,
+      "step": 23311
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.931234251633218e-06,
+      "loss": 0.0001,
+      "step": 23312
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.92813479713686e-06,
+      "loss": 0.0008,
+      "step": 23313
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.925037815720143e-06,
+      "loss": 0.0005,
+      "step": 23314
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.921943307434809e-06,
+      "loss": 0.0001,
+      "step": 23315
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.91885127233255e-06,
+      "loss": 0.0041,
+      "step": 23316
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.915761710464991e-06,
+      "loss": 0.0058,
+      "step": 23317
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9126746218837417e-06,
+      "loss": 0.0001,
+      "step": 23318
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9095900066403436e-06,
+      "loss": 0.0106,
+      "step": 23319
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9065078647863551e-06,
+      "loss": 0.0019,
+      "step": 23320
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9034281963732189e-06,
+      "loss": 0.0001,
+      "step": 23321
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9003510014523938e-06,
+      "loss": 0.0001,
+      "step": 23322
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.897276280075255e-06,
+      "loss": 0.0005,
+      "step": 23323
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8942040322931785e-06,
+      "loss": 0.0045,
+      "step": 23324
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8911342581574562e-06,
+      "loss": 0.0092,
+      "step": 23325
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8880669577193807e-06,
+      "loss": 0.0017,
+      "step": 23326
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8850021310301444e-06,
+      "loss": 0.0018,
+      "step": 23327
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8819397781409563e-06,
+      "loss": 0.0029,
+      "step": 23328
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.878879899102992e-06,
+      "loss": 0.0055,
+      "step": 23329
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8758224939673106e-06,
+      "loss": 0.0083,
+      "step": 23330
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8727675627850214e-06,
+      "loss": 0.0006,
+      "step": 23331
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8697151056071002e-06,
+      "loss": 0.0004,
+      "step": 23332
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8666651224845731e-06,
+      "loss": 0.0014,
+      "step": 23333
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.863617613468349e-06,
+      "loss": 0.0002,
+      "step": 23334
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8605725786093374e-06,
+      "loss": 0.0015,
+      "step": 23335
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8575300179583974e-06,
+      "loss": 0.0003,
+      "step": 23336
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8544899315663387e-06,
+      "loss": 0.0001,
+      "step": 23337
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8514523194839537e-06,
+      "loss": 0.0118,
+      "step": 23338
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.848417181761952e-06,
+      "loss": 0.0007,
+      "step": 23339
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.845384518451043e-06,
+      "loss": 0.0007,
+      "step": 23340
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8423543296018861e-06,
+      "loss": 0.0024,
+      "step": 23341
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8393266152650743e-06,
+      "loss": 0.0005,
+      "step": 23342
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.836301375491167e-06,
+      "loss": 0.0009,
+      "step": 23343
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.833278610330724e-06,
+      "loss": 0.0002,
+      "step": 23344
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8302583198342046e-06,
+      "loss": 0.0013,
+      "step": 23345
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8272405040520688e-06,
+      "loss": 0.0002,
+      "step": 23346
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8242251630346927e-06,
+      "loss": 0.0026,
+      "step": 23347
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.821212296832486e-06,
+      "loss": 0.004,
+      "step": 23348
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8182019054957252e-06,
+      "loss": 0.002,
+      "step": 23349
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.815193989074687e-06,
+      "loss": 0.0018,
+      "step": 23350
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.812188547619664e-06,
+      "loss": 0.0009,
+      "step": 23351
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8091855811807832e-06,
+      "loss": 0.0001,
+      "step": 23352
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8061850898082708e-06,
+      "loss": 0.0007,
+      "step": 23353
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8031870735521869e-06,
+      "loss": 0.0017,
+      "step": 23354
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8001915324626082e-06,
+      "loss": 0.0011,
+      "step": 23355
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7971984665895778e-06,
+      "loss": 0.003,
+      "step": 23356
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7942078759831058e-06,
+      "loss": 0.0003,
+      "step": 23357
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7912197606931023e-06,
+      "loss": 0.0017,
+      "step": 23358
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7882341207694938e-06,
+      "loss": 0.0021,
+      "step": 23359
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7852509562621576e-06,
+      "loss": 0.0001,
+      "step": 23360
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7822702672208866e-06,
+      "loss": 0.0021,
+      "step": 23361
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.779292053695508e-06,
+      "loss": 0.0004,
+      "step": 23362
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7763163157357153e-06,
+      "loss": 0.0016,
+      "step": 23363
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7733430533912518e-06,
+      "loss": 0.0002,
+      "step": 23364
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7703722667117281e-06,
+      "loss": 0.0037,
+      "step": 23365
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7674039557468212e-06,
+      "loss": 0.0024,
+      "step": 23366
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7644381205460745e-06,
+      "loss": 0.0007,
+      "step": 23367
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7614747611590152e-06,
+      "loss": 0.0038,
+      "step": 23368
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.758513877635137e-06,
+      "loss": 0.0016,
+      "step": 23369
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7555554700239173e-06,
+      "loss": 0.0002,
+      "step": 23370
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7525995383747494e-06,
+      "loss": 0.0007,
+      "step": 23371
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7496460827370106e-06,
+      "loss": 0.0006,
+      "step": 23372
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7466951031600118e-06,
+      "loss": 0.0009,
+      "step": 23373
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7437465996930632e-06,
+      "loss": 0.0019,
+      "step": 23374
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7408005723853925e-06,
+      "loss": 0.0017,
+      "step": 23375
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7378570212862264e-06,
+      "loss": 0.0046,
+      "step": 23376
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7349159464447094e-06,
+      "loss": 0.0009,
+      "step": 23377
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7319773479099518e-06,
+      "loss": 0.0023,
+      "step": 23378
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7290412257310648e-06,
+      "loss": 0.0039,
+      "step": 23379
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7261075799570424e-06,
+      "loss": 0.0078,
+      "step": 23380
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7231764106369284e-06,
+      "loss": 0.0003,
+      "step": 23381
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7202477178196838e-06,
+      "loss": 0.0,
+      "step": 23382
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7173215015541531e-06,
+      "loss": 0.0017,
+      "step": 23383
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7143977618892968e-06,
+      "loss": 0.0035,
+      "step": 23384
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7114764988738927e-06,
+      "loss": 0.0003,
+      "step": 23385
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7085577125567517e-06,
+      "loss": 0.0001,
+      "step": 23386
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7056414029866016e-06,
+      "loss": 0.0034,
+      "step": 23387
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7027275702121702e-06,
+      "loss": 0.0013,
+      "step": 23388
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.699816214282118e-06,
+      "loss": 0.0002,
+      "step": 23389
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.69690733524509e-06,
+      "loss": 0.0015,
+      "step": 23390
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6940009331496308e-06,
+      "loss": 0.0033,
+      "step": 23391
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6910970080443008e-06,
+      "loss": 0.0001,
+      "step": 23392
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6881955599776288e-06,
+      "loss": 0.0114,
+      "step": 23393
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6852965889980251e-06,
+      "loss": 0.0001,
+      "step": 23394
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6824000951539518e-06,
+      "loss": 0.0001,
+      "step": 23395
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6795060784937363e-06,
+      "loss": 0.0001,
+      "step": 23396
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6766145390657737e-06,
+      "loss": 0.0019,
+      "step": 23397
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6737254769183085e-06,
+      "loss": 0.0019,
+      "step": 23398
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6708388920996352e-06,
+      "loss": 0.0024,
+      "step": 23399
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6679547846579323e-06,
+      "loss": 0.0015,
+      "step": 23400
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6650731546413776e-06,
+      "loss": 0.0054,
+      "step": 23401
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6621940020980995e-06,
+      "loss": 0.0019,
+      "step": 23402
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6593173270761927e-06,
+      "loss": 0.0012,
+      "step": 23403
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6564431296237023e-06,
+      "loss": 0.0019,
+      "step": 23404
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6535714097886399e-06,
+      "loss": 0.003,
+      "step": 23405
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6507021676189503e-06,
+      "loss": 0.0026,
+      "step": 23406
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6478354031625451e-06,
+      "loss": 0.001,
+      "step": 23407
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.644971116467353e-06,
+      "loss": 0.0006,
+      "step": 23408
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6421093075811687e-06,
+      "loss": 0.0021,
+      "step": 23409
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6392499765518042e-06,
+      "loss": 0.0004,
+      "step": 23410
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6363931234270044e-06,
+      "loss": 0.0003,
+      "step": 23411
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6335387482545148e-06,
+      "loss": 0.0001,
+      "step": 23412
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.630686851081997e-06,
+      "loss": 0.0003,
+      "step": 23413
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6278374319570465e-06,
+      "loss": 0.0015,
+      "step": 23414
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.624990490927308e-06,
+      "loss": 0.0003,
+      "step": 23415
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6221460280402942e-06,
+      "loss": 0.0003,
+      "step": 23416
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6193040433435333e-06,
+      "loss": 0.0007,
+      "step": 23417
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.616464536884471e-06,
+      "loss": 0.0023,
+      "step": 23418
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6136275087105522e-06,
+      "loss": 0.0014,
+      "step": 23419
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.610792958869156e-06,
+      "loss": 0.0001,
+      "step": 23420
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.607960887407611e-06,
+      "loss": 0.0007,
+      "step": 23421
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6051312943732297e-06,
+      "loss": 0.0006,
+      "step": 23422
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6023041798132741e-06,
+      "loss": 0.0001,
+      "step": 23423
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.599479543774973e-06,
+      "loss": 0.0003,
+      "step": 23424
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5966573863054722e-06,
+      "loss": 0.0004,
+      "step": 23425
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5938377074519338e-06,
+      "loss": 0.0037,
+      "step": 23426
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5910205072614368e-06,
+      "loss": 0.0008,
+      "step": 23427
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5882057857810438e-06,
+      "loss": 0.0002,
+      "step": 23428
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5853935430577501e-06,
+      "loss": 0.0044,
+      "step": 23429
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.582583779138552e-06,
+      "loss": 0.0007,
+      "step": 23430
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5797764940703617e-06,
+      "loss": 0.0064,
+      "step": 23431
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5769716879000583e-06,
+      "loss": 0.0004,
+      "step": 23432
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5741693606745043e-06,
+      "loss": 0.0005,
+      "step": 23433
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.571369512440479e-06,
+      "loss": 0.0022,
+      "step": 23434
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5685721432447785e-06,
+      "loss": 0.0019,
+      "step": 23435
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5657772531340983e-06,
+      "loss": 0.0016,
+      "step": 23436
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.562984842155135e-06,
+      "loss": 0.0003,
+      "step": 23437
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.560194910354501e-06,
+      "loss": 0.0001,
+      "step": 23438
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5574074577788254e-06,
+      "loss": 0.0009,
+      "step": 23439
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5546224844746548e-06,
+      "loss": 0.0001,
+      "step": 23440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5518399904885014e-06,
+      "loss": 0.0014,
+      "step": 23441
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5490599758668286e-06,
+      "loss": 0.0001,
+      "step": 23442
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.546282440656066e-06,
+      "loss": 0.0002,
+      "step": 23443
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5435073849026259e-06,
+      "loss": 0.0004,
+      "step": 23444
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5407348086528381e-06,
+      "loss": 0.0003,
+      "step": 23445
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5379647119530325e-06,
+      "loss": 0.0012,
+      "step": 23446
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5351970948494385e-06,
+      "loss": 0.0022,
+      "step": 23447
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5324319573883192e-06,
+      "loss": 0.0001,
+      "step": 23448
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5296692996158045e-06,
+      "loss": 0.0008,
+      "step": 23449
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.526909121578107e-06,
+      "loss": 0.0006,
+      "step": 23450
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5241514233212737e-06,
+      "loss": 0.0031,
+      "step": 23451
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5213962048913842e-06,
+      "loss": 0.0013,
+      "step": 23452
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5186434663344515e-06,
+      "loss": 0.0014,
+      "step": 23453
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5158932076964392e-06,
+      "loss": 0.0011,
+      "step": 23454
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5131454290233102e-06,
+      "loss": 0.0006,
+      "step": 23455
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5104001303609282e-06,
+      "loss": 0.0013,
+      "step": 23456
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.507657311755156e-06,
+      "loss": 0.0006,
+      "step": 23457
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5049169732518075e-06,
+      "loss": 0.0043,
+      "step": 23458
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5021791148966789e-06,
+      "loss": 0.0002,
+      "step": 23459
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4994437367354339e-06,
+      "loss": 0.0014,
+      "step": 23460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4967108388138027e-06,
+      "loss": 0.0002,
+      "step": 23461
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4939804211774153e-06,
+      "loss": 0.0001,
+      "step": 23462
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4912524838718852e-06,
+      "loss": 0.0023,
+      "step": 23463
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4885270269427763e-06,
+      "loss": 0.0018,
+      "step": 23464
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4858040504355851e-06,
+      "loss": 0.0004,
+      "step": 23465
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4830835543958254e-06,
+      "loss": 0.0009,
+      "step": 23466
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.480365538868894e-06,
+      "loss": 0.0028,
+      "step": 23467
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4776500039002381e-06,
+      "loss": 0.0022,
+      "step": 23468
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4749369495351715e-06,
+      "loss": 0.0007,
+      "step": 23469
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4722263758190077e-06,
+      "loss": 0.001,
+      "step": 23470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4695182827970276e-06,
+      "loss": 0.0006,
+      "step": 23471
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4668126705144777e-06,
+      "loss": 0.0014,
+      "step": 23472
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4641095390165225e-06,
+      "loss": 0.0036,
+      "step": 23473
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4614088883483255e-06,
+      "loss": 0.0004,
+      "step": 23474
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.458710718554984e-06,
+      "loss": 0.0003,
+      "step": 23475
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.456015029681562e-06,
+      "loss": 0.0005,
+      "step": 23476
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4533218217730901e-06,
+      "loss": 0.001,
+      "step": 23477
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4506310948745493e-06,
+      "loss": 0.0005,
+      "step": 23478
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4479428490308697e-06,
+      "loss": 0.0005,
+      "step": 23479
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4452570842869493e-06,
+      "loss": 0.001,
+      "step": 23480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4425738006876686e-06,
+      "loss": 0.0026,
+      "step": 23481
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.439892998277825e-06,
+      "loss": 0.0016,
+      "step": 23482
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.437214677102183e-06,
+      "loss": 0.0055,
+      "step": 23483
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4345388372055067e-06,
+      "loss": 0.0,
+      "step": 23484
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4318654786324434e-06,
+      "loss": 0.0012,
+      "step": 23485
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.429194601427691e-06,
+      "loss": 0.0005,
+      "step": 23486
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4265262056358307e-06,
+      "loss": 0.0049,
+      "step": 23487
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4238602913014263e-06,
+      "loss": 0.0006,
+      "step": 23488
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4211968584690092e-06,
+      "loss": 0.0015,
+      "step": 23489
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.418535907183077e-06,
+      "loss": 0.0024,
+      "step": 23490
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.415877437488061e-06,
+      "loss": 0.0005,
+      "step": 23491
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4132214494283755e-06,
+      "loss": 0.001,
+      "step": 23492
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4105679430483518e-06,
+      "loss": 0.0003,
+      "step": 23493
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4079169183923377e-06,
+      "loss": 0.0018,
+      "step": 23494
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4052683755045812e-06,
+      "loss": 0.0002,
+      "step": 23495
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.402622314429347e-06,
+      "loss": 0.0011,
+      "step": 23496
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3999787352108326e-06,
+      "loss": 0.0001,
+      "step": 23497
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3973376378931533e-06,
+      "loss": 0.0052,
+      "step": 23498
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3946990225204569e-06,
+      "loss": 0.0001,
+      "step": 23499
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3920628891367913e-06,
+      "loss": 0.0027,
+      "step": 23500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3894292377861883e-06,
+      "loss": 0.001,
+      "step": 23501
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3867980685126457e-06,
+      "loss": 0.0003,
+      "step": 23502
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3841693813601118e-06,
+      "loss": 0.001,
+      "step": 23503
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3815431763724683e-06,
+      "loss": 0.0002,
+      "step": 23504
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3789194535935967e-06,
+      "loss": 0.0008,
+      "step": 23505
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3762982130673116e-06,
+      "loss": 0.0078,
+      "step": 23506
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3736794548373953e-06,
+      "loss": 0.0002,
+      "step": 23507
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.371063178947579e-06,
+      "loss": 0.0005,
+      "step": 23508
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3684493854415612e-06,
+      "loss": 0.0012,
+      "step": 23509
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3658380743630071e-06,
+      "loss": 0.0017,
+      "step": 23510
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3632292457555149e-06,
+      "loss": 0.0023,
+      "step": 23511
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3606228996626833e-06,
+      "loss": 0.0038,
+      "step": 23512
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3580190361280108e-06,
+      "loss": 0.0001,
+      "step": 23513
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3554176551950124e-06,
+      "loss": 0.0009,
+      "step": 23514
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3528187569071202e-06,
+      "loss": 0.0001,
+      "step": 23515
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3502223413077496e-06,
+      "loss": 0.0002,
+      "step": 23516
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3476284084402654e-06,
+      "loss": 0.0067,
+      "step": 23517
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.345036958347967e-06,
+      "loss": 0.0019,
+      "step": 23518
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3424479910741859e-06,
+      "loss": 0.0001,
+      "step": 23519
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.339861506662121e-06,
+      "loss": 0.0011,
+      "step": 23520
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3372775051549877e-06,
+      "loss": 0.0004,
+      "step": 23521
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.334695986595935e-06,
+      "loss": 0.0007,
+      "step": 23522
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.332116951028095e-06,
+      "loss": 0.0002,
+      "step": 23523
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3295403984945164e-06,
+      "loss": 0.0009,
+      "step": 23524
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3269663290382648e-06,
+      "loss": 0.0053,
+      "step": 23525
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.324394742702306e-06,
+      "loss": 0.0016,
+      "step": 23526
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3218256395296056e-06,
+      "loss": 0.0002,
+      "step": 23527
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3192590195630625e-06,
+      "loss": 0.0034,
+      "step": 23528
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3166948828455425e-06,
+      "loss": 0.0015,
+      "step": 23529
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3141332294198948e-06,
+      "loss": 0.002,
+      "step": 23530
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3115740593288683e-06,
+      "loss": 0.0025,
+      "step": 23531
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3090173726152453e-06,
+      "loss": 0.0003,
+      "step": 23532
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3064631693216921e-06,
+      "loss": 0.0002,
+      "step": 23533
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3039114494908908e-06,
+      "loss": 0.0022,
+      "step": 23534
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3013622131654412e-06,
+      "loss": 0.0038,
+      "step": 23535
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2988154603879419e-06,
+      "loss": 0.0025,
+      "step": 23536
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2962711912009094e-06,
+      "loss": 0.0023,
+      "step": 23537
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.293729405646843e-06,
+      "loss": 0.0002,
+      "step": 23538
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2911901037682083e-06,
+      "loss": 0.0035,
+      "step": 23539
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.288653285607405e-06,
+      "loss": 0.0004,
+      "step": 23540
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2861189512067994e-06,
+      "loss": 0.0037,
+      "step": 23541
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.283587100608724e-06,
+      "loss": 0.005,
+      "step": 23542
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2810577338554783e-06,
+      "loss": 0.0001,
+      "step": 23543
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2785308509892788e-06,
+      "loss": 0.0003,
+      "step": 23544
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2760064520523583e-06,
+      "loss": 0.0008,
+      "step": 23545
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2734845370868496e-06,
+      "loss": 0.0025,
+      "step": 23546
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2709651061349025e-06,
+      "loss": 0.0009,
+      "step": 23547
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2684481592385998e-06,
+      "loss": 0.0005,
+      "step": 23548
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2659336964399414e-06,
+      "loss": 0.0045,
+      "step": 23549
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2634217177809603e-06,
+      "loss": 0.0004,
+      "step": 23550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2609122233035728e-06,
+      "loss": 0.0001,
+      "step": 23551
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2584052130497457e-06,
+      "loss": 0.0004,
+      "step": 23552
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.255900687061312e-06,
+      "loss": 0.0014,
+      "step": 23553
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2533986453801048e-06,
+      "loss": 0.0003,
+      "step": 23554
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2508990880479076e-06,
+      "loss": 0.0049,
+      "step": 23555
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2484020151065034e-06,
+      "loss": 0.0036,
+      "step": 23556
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.245907426597559e-06,
+      "loss": 0.0007,
+      "step": 23557
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2434153225627574e-06,
+      "loss": 0.0041,
+      "step": 23558
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2409257030437158e-06,
+      "loss": 0.0035,
+      "step": 23559
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2384385680820174e-06,
+      "loss": 0.0002,
+      "step": 23560
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.235953917719229e-06,
+      "loss": 0.0003,
+      "step": 23561
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.233471751996784e-06,
+      "loss": 0.0054,
+      "step": 23562
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2309920709561994e-06,
+      "loss": 0.0022,
+      "step": 23563
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2285148746388584e-06,
+      "loss": 0.0008,
+      "step": 23564
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.226040163086145e-06,
+      "loss": 0.0002,
+      "step": 23565
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2235679363393925e-06,
+      "loss": 0.0002,
+      "step": 23566
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2210981944399011e-06,
+      "loss": 0.0001,
+      "step": 23567
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.218630937428905e-06,
+      "loss": 0.0008,
+      "step": 23568
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.216166165347604e-06,
+      "loss": 0.0001,
+      "step": 23569
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2137038782371989e-06,
+      "loss": 0.0019,
+      "step": 23570
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2112440761387732e-06,
+      "loss": 0.0055,
+      "step": 23571
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2087867590934442e-06,
+      "loss": 0.0003,
+      "step": 23572
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.206331927142229e-06,
+      "loss": 0.002,
+      "step": 23573
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2038795803261448e-06,
+      "loss": 0.0011,
+      "step": 23574
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.201429718686142e-06,
+      "loss": 0.0001,
+      "step": 23575
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1989823422631384e-06,
+      "loss": 0.0025,
+      "step": 23576
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1965374510980175e-06,
+      "loss": 0.0006,
+      "step": 23577
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1940950452315967e-06,
+      "loss": 0.0019,
+      "step": 23578
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1916551247046935e-06,
+      "loss": 0.0001,
+      "step": 23579
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.189217689558042e-06,
+      "loss": 0.0025,
+      "step": 23580
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1867827398323593e-06,
+      "loss": 0.0004,
+      "step": 23581
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1843502755682966e-06,
+      "loss": 0.0015,
+      "step": 23582
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1819202968065045e-06,
+      "loss": 0.0043,
+      "step": 23583
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1794928035875507e-06,
+      "loss": 0.0029,
+      "step": 23584
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.177067795951986e-06,
+      "loss": 0.0002,
+      "step": 23585
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1746452739403111e-06,
+      "loss": 0.0022,
+      "step": 23586
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1722252375929942e-06,
+      "loss": 0.0002,
+      "step": 23587
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.169807686950436e-06,
+      "loss": 0.0006,
+      "step": 23588
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1673926220530371e-06,
+      "loss": 0.0014,
+      "step": 23589
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1649800429411327e-06,
+      "loss": 0.0003,
+      "step": 23590
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.16256994965499e-06,
+      "loss": 0.0013,
+      "step": 23591
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.160162342234877e-06,
+      "loss": 0.0001,
+      "step": 23592
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1577572207210283e-06,
+      "loss": 0.0031,
+      "step": 23593
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1553545851535784e-06,
+      "loss": 0.0004,
+      "step": 23594
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1529544355726783e-06,
+      "loss": 0.0005,
+      "step": 23595
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1505567720184127e-06,
+      "loss": 0.0039,
+      "step": 23596
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1481615945307999e-06,
+      "loss": 0.0033,
+      "step": 23597
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1457689031498906e-06,
+      "loss": 0.0007,
+      "step": 23598
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1433786979156035e-06,
+      "loss": 0.0001,
+      "step": 23599
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1409909788678895e-06,
+      "loss": 0.0001,
+      "step": 23600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.138605746046617e-06,
+      "loss": 0.0005,
+      "step": 23601
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1362229994916206e-06,
+      "loss": 0.001,
+      "step": 23602
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1338427392427018e-06,
+      "loss": 0.0028,
+      "step": 23603
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1314649653396123e-06,
+      "loss": 0.0053,
+      "step": 23604
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1290896778220703e-06,
+      "loss": 0.0052,
+      "step": 23605
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.126716876729744e-06,
+      "loss": 0.0045,
+      "step": 23606
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1243465621022685e-06,
+      "loss": 0.0023,
+      "step": 23607
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1219787339792286e-06,
+      "loss": 0.0003,
+      "step": 23608
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1196133924001592e-06,
+      "loss": 0.0022,
+      "step": 23609
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1172505374045793e-06,
+      "loss": 0.0019,
+      "step": 23610
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1148901690319568e-06,
+      "loss": 0.0017,
+      "step": 23611
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1125322873217269e-06,
+      "loss": 0.0048,
+      "step": 23612
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1101768923132247e-06,
+      "loss": 0.0003,
+      "step": 23613
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1078239840458359e-06,
+      "loss": 0.0011,
+      "step": 23614
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1054735625588284e-06,
+      "loss": 0.0021,
+      "step": 23615
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1031256278914712e-06,
+      "loss": 0.0027,
+      "step": 23616
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.100780180082983e-06,
+      "loss": 0.0001,
+      "step": 23617
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.098437219172532e-06,
+      "loss": 0.0001,
+      "step": 23618
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0960967451992543e-06,
+      "loss": 0.0012,
+      "step": 23619
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0937587582022345e-06,
+      "loss": 0.0005,
+      "step": 23620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0914232582205252e-06,
+      "loss": 0.0004,
+      "step": 23621
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0890902452931282e-06,
+      "loss": 0.0001,
+      "step": 23622
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0867597194590127e-06,
+      "loss": 0.0027,
+      "step": 23623
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0844316807571141e-06,
+      "loss": 0.0003,
+      "step": 23624
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.082106129226301e-06,
+      "loss": 0.0031,
+      "step": 23625
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0797830649054096e-06,
+      "loss": 0.0001,
+      "step": 23626
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0774624878332583e-06,
+      "loss": 0.0014,
+      "step": 23627
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.075144398048583e-06,
+      "loss": 0.0011,
+      "step": 23628
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0728287955901195e-06,
+      "loss": 0.0002,
+      "step": 23629
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0705156804965365e-06,
+      "loss": 0.001,
+      "step": 23630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0682050528064534e-06,
+      "loss": 0.0015,
+      "step": 23631
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0658969125584893e-06,
+      "loss": 0.0038,
+      "step": 23632
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0635912597911633e-06,
+      "loss": 0.0023,
+      "step": 23633
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.061288094543028e-06,
+      "loss": 0.0024,
+      "step": 23634
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0589874168525025e-06,
+      "loss": 0.0013,
+      "step": 23635
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0566892267580395e-06,
+      "loss": 0.0005,
+      "step": 23636
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0543935242980085e-06,
+      "loss": 0.001,
+      "step": 23637
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0521003095107782e-06,
+      "loss": 0.0068,
+      "step": 23638
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0498095824346187e-06,
+      "loss": 0.0013,
+      "step": 23639
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0475213431077989e-06,
+      "loss": 0.0033,
+      "step": 23640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.045235591568555e-06,
+      "loss": 0.0009,
+      "step": 23641
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.04295232785504e-06,
+      "loss": 0.0013,
+      "step": 23642
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0406715520053899e-06,
+      "loss": 0.0025,
+      "step": 23643
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0383932640577075e-06,
+      "loss": 0.0,
+      "step": 23644
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0361174640500626e-06,
+      "loss": 0.0001,
+      "step": 23645
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0338441520204243e-06,
+      "loss": 0.0077,
+      "step": 23646
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0315733280067962e-06,
+      "loss": 0.0035,
+      "step": 23647
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0293049920470643e-06,
+      "loss": 0.0,
+      "step": 23648
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0270391441791813e-06,
+      "loss": 0.0022,
+      "step": 23649
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0247757844409177e-06,
+      "loss": 0.0017,
+      "step": 23650
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.022514912870126e-06,
+      "loss": 0.0009,
+      "step": 23651
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0202565295045428e-06,
+      "loss": 0.0008,
+      "step": 23652
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0180006343819047e-06,
+      "loss": 0.0022,
+      "step": 23653
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0157472275398649e-06,
+      "loss": 0.0012,
+      "step": 23654
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.013496309016093e-06,
+      "loss": 0.0006,
+      "step": 23655
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0112478788481425e-06,
+      "loss": 0.0007,
+      "step": 23656
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0090019370736003e-06,
+      "loss": 0.0032,
+      "step": 23657
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.006758483729969e-06,
+      "loss": 0.0011,
+      "step": 23658
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0045175188547194e-06,
+      "loss": 0.0002,
+      "step": 23659
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0022790424852545e-06,
+      "loss": 0.0014,
+      "step": 23660
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.000043054659011e-06,
+      "loss": 0.0004,
+      "step": 23661
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.978095554132759e-07,
+      "loss": 0.0025,
+      "step": 23662
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.955785447854193e-07,
+      "loss": 0.0015,
+      "step": 23663
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.933500228126278e-07,
+      "loss": 0.0036,
+      "step": 23664
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.911239895321887e-07,
+      "loss": 0.0012,
+      "step": 23665
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.889004449812388e-07,
+      "loss": 0.0019,
+      "step": 23666
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.866793891969482e-07,
+      "loss": 0.0007,
+      "step": 23667
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.844608222163708e-07,
+      "loss": 0.0001,
+      "step": 23668
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.822447440765936e-07,
+      "loss": 0.0009,
+      "step": 23669
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.8003115481462e-07,
+      "loss": 0.0004,
+      "step": 23670
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.778200544674208e-07,
+      "loss": 0.0031,
+      "step": 23671
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.756114430719331e-07,
+      "loss": 0.0,
+      "step": 23672
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.734053206650273e-07,
+      "loss": 0.0007,
+      "step": 23673
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.712016872835737e-07,
+      "loss": 0.0007,
+      "step": 23674
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.6900054296436e-07,
+      "loss": 0.0001,
+      "step": 23675
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.668018877441398e-07,
+      "loss": 0.0004,
+      "step": 23676
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.646057216596504e-07,
+      "loss": 0.0001,
+      "step": 23677
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.624120447475959e-07,
+      "loss": 0.0145,
+      "step": 23678
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.602208570445636e-07,
+      "loss": 0.009,
+      "step": 23679
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.580321585871742e-07,
+      "loss": 0.0016,
+      "step": 23680
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.558459494119985e-07,
+      "loss": 0.0008,
+      "step": 23681
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.536622295555241e-07,
+      "loss": 0.002,
+      "step": 23682
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.514809990542383e-07,
+      "loss": 0.0075,
+      "step": 23683
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.493022579445619e-07,
+      "loss": 0.0016,
+      "step": 23684
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.471260062629161e-07,
+      "loss": 0.0001,
+      "step": 23685
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.449522440456047e-07,
+      "loss": 0.0008,
+      "step": 23686
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.427809713289492e-07,
+      "loss": 0.0006,
+      "step": 23687
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.406121881492201e-07,
+      "loss": 0.0013,
+      "step": 23688
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.384458945426388e-07,
+      "loss": 0.0032,
+      "step": 23689
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.362820905453927e-07,
+      "loss": 0.0017,
+      "step": 23690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.341207761936031e-07,
+      "loss": 0.0046,
+      "step": 23691
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.319619515233745e-07,
+      "loss": 0.0015,
+      "step": 23692
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.298056165707779e-07,
+      "loss": 0.0005,
+      "step": 23693
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.276517713718013e-07,
+      "loss": 0.0036,
+      "step": 23694
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.25500415962449e-07,
+      "loss": 0.0061,
+      "step": 23695
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.233515503786426e-07,
+      "loss": 0.0009,
+      "step": 23696
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.212051746562532e-07,
+      "loss": 0.0021,
+      "step": 23697
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.190612888311521e-07,
+      "loss": 0.0006,
+      "step": 23698
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.169198929391275e-07,
+      "loss": 0.0,
+      "step": 23699
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.147809870159673e-07,
+      "loss": 0.0027,
+      "step": 23700
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.126445710973929e-07,
+      "loss": 0.0005,
+      "step": 23701
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.105106452190591e-07,
+      "loss": 0.0038,
+      "step": 23702
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.083792094166209e-07,
+      "loss": 0.0006,
+      "step": 23703
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.06250263725683e-07,
+      "loss": 0.0004,
+      "step": 23704
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.041238081818003e-07,
+      "loss": 0.0001,
+      "step": 23705
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.019998428204944e-07,
+      "loss": 0.0001,
+      "step": 23706
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.99878367677237e-07,
+      "loss": 0.0003,
+      "step": 23707
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.977593827874496e-07,
+      "loss": 0.0031,
+      "step": 23708
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.956428881865374e-07,
+      "loss": 0.0,
+      "step": 23709
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.935288839098221e-07,
+      "loss": 0.0025,
+      "step": 23710
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.914173699926585e-07,
+      "loss": 0.0031,
+      "step": 23711
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.893083464702689e-07,
+      "loss": 0.0041,
+      "step": 23712
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.87201813377908e-07,
+      "loss": 0.0005,
+      "step": 23713
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.850977707507478e-07,
+      "loss": 0.0002,
+      "step": 23714
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.82996218623927e-07,
+      "loss": 0.0005,
+      "step": 23715
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.808971570325507e-07,
+      "loss": 0.0032,
+      "step": 23716
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.788005860116576e-07,
+      "loss": 0.0011,
+      "step": 23717
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.767065055963029e-07,
+      "loss": 0.0107,
+      "step": 23718
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.746149158214255e-07,
+      "loss": 0.0044,
+      "step": 23719
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.725258167219807e-07,
+      "loss": 0.0009,
+      "step": 23720
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.704392083328404e-07,
+      "loss": 0.0,
+      "step": 23721
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.68355090688877e-07,
+      "loss": 0.0021,
+      "step": 23722
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.662734638248958e-07,
+      "loss": 0.0028,
+      "step": 23723
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.641943277756524e-07,
+      "loss": 0.0009,
+      "step": 23724
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.621176825758691e-07,
+      "loss": 0.0011,
+      "step": 23725
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.600435282602347e-07,
+      "loss": 0.0001,
+      "step": 23726
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.579718648634215e-07,
+      "loss": 0.0002,
+      "step": 23727
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.559026924199853e-07,
+      "loss": 0.0001,
+      "step": 23728
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.538360109644982e-07,
+      "loss": 0.0026,
+      "step": 23729
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.517718205314828e-07,
+      "loss": 0.0001,
+      "step": 23730
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.497101211554281e-07,
+      "loss": 0.0045,
+      "step": 23731
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.476509128707565e-07,
+      "loss": 0.0003,
+      "step": 23732
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.455941957118573e-07,
+      "loss": 0.0008,
+      "step": 23733
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.435399697130696e-07,
+      "loss": 0.0028,
+      "step": 23734
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.414882349087326e-07,
+      "loss": 0.001,
+      "step": 23735
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.394389913330857e-07,
+      "loss": 0.0004,
+      "step": 23736
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.373922390203846e-07,
+      "loss": 0.0021,
+      "step": 23737
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.353479780047856e-07,
+      "loss": 0.0003,
+      "step": 23738
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.333062083204445e-07,
+      "loss": 0.0015,
+      "step": 23739
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.312669300014674e-07,
+      "loss": 0.0017,
+      "step": 23740
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.292301430819104e-07,
+      "loss": 0.0012,
+      "step": 23741
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.271958475957796e-07,
+      "loss": 0.0003,
+      "step": 23742
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.251640435770646e-07,
+      "loss": 0.0024,
+      "step": 23743
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.231347310597047e-07,
+      "loss": 0.0006,
+      "step": 23744
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.21107910077573e-07,
+      "loss": 0.0012,
+      "step": 23745
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.190835806645424e-07,
+      "loss": 0.0063,
+      "step": 23746
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.170617428544191e-07,
+      "loss": 0.0006,
+      "step": 23747
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.150423966809594e-07,
+      "loss": 0.006,
+      "step": 23748
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.130255421779031e-07,
+      "loss": 0.0018,
+      "step": 23749
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.110111793789231e-07,
+      "loss": 0.0012,
+      "step": 23750
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.089993083176927e-07,
+      "loss": 0.0015,
+      "step": 23751
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.069899290277681e-07,
+      "loss": 0.0073,
+      "step": 23752
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.049830415427395e-07,
+      "loss": 0.0002,
+      "step": 23753
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.029786458961296e-07,
+      "loss": 0.0028,
+      "step": 23754
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.009767421213953e-07,
+      "loss": 0.0009,
+      "step": 23755
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.989773302519931e-07,
+      "loss": 0.002,
+      "step": 23756
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.969804103212963e-07,
+      "loss": 0.0001,
+      "step": 23757
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.949859823626614e-07,
+      "loss": 0.0038,
+      "step": 23758
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.92994046409412e-07,
+      "loss": 0.002,
+      "step": 23759
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.910046024948213e-07,
+      "loss": 0.0002,
+      "step": 23760
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.890176506520796e-07,
+      "loss": 0.0009,
+      "step": 23761
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.870331909144268e-07,
+      "loss": 0.0005,
+      "step": 23762
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.850512233149697e-07,
+      "loss": 0.0001,
+      "step": 23763
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.830717478868154e-07,
+      "loss": 0.0035,
+      "step": 23764
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.810947646630373e-07,
+      "loss": 0.0004,
+      "step": 23765
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.791202736766255e-07,
+      "loss": 0.0032,
+      "step": 23766
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.771482749606039e-07,
+      "loss": 0.0035,
+      "step": 23767
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.751787685478794e-07,
+      "loss": 0.0018,
+      "step": 23768
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.732117544713423e-07,
+      "loss": 0.0019,
+      "step": 23769
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.712472327638663e-07,
+      "loss": 0.0002,
+      "step": 23770
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.69285203458242e-07,
+      "loss": 0.0009,
+      "step": 23771
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.673256665872596e-07,
+      "loss": 0.0003,
+      "step": 23772
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.653686221836264e-07,
+      "loss": 0.0002,
+      "step": 23773
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.634140702800329e-07,
+      "loss": 0.0001,
+      "step": 23774
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.614620109091363e-07,
+      "loss": 0.0009,
+      "step": 23775
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.595124441035271e-07,
+      "loss": 0.0066,
+      "step": 23776
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.575653698957629e-07,
+      "loss": 0.0027,
+      "step": 23777
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.556207883183841e-07,
+      "loss": 0.0019,
+      "step": 23778
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.536786994038479e-07,
+      "loss": 0.0006,
+      "step": 23779
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.51739103184612e-07,
+      "loss": 0.0003,
+      "step": 23780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.498019996930505e-07,
+      "loss": 0.0006,
+      "step": 23781
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.478673889615372e-07,
+      "loss": 0.0009,
+      "step": 23782
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.459352710223631e-07,
+      "loss": 0.0009,
+      "step": 23783
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.440056459078025e-07,
+      "loss": 0.0088,
+      "step": 23784
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.420785136500961e-07,
+      "loss": 0.0096,
+      "step": 23785
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.401538742814183e-07,
+      "loss": 0.0005,
+      "step": 23786
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.382317278339267e-07,
+      "loss": 0.0011,
+      "step": 23787
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.363120743397122e-07,
+      "loss": 0.0009,
+      "step": 23788
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.343949138308325e-07,
+      "loss": 0.0,
+      "step": 23789
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.324802463393287e-07,
+      "loss": 0.0009,
+      "step": 23790
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.305680718971585e-07,
+      "loss": 0.0023,
+      "step": 23791
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.286583905362631e-07,
+      "loss": 0.0032,
+      "step": 23792
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.267512022885502e-07,
+      "loss": 0.0045,
+      "step": 23793
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.248465071858611e-07,
+      "loss": 0.0012,
+      "step": 23794
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.229443052600204e-07,
+      "loss": 0.0047,
+      "step": 23795
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.210445965427858e-07,
+      "loss": 0.0037,
+      "step": 23796
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.191473810658821e-07,
+      "loss": 0.0003,
+      "step": 23797
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.172526588610339e-07,
+      "loss": 0.0,
+      "step": 23798
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.153604299598159e-07,
+      "loss": 0.0006,
+      "step": 23799
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.134706943939029e-07,
+      "loss": 0.0001,
+      "step": 23800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.115834521948194e-07,
+      "loss": 0.0018,
+      "step": 23801
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.096987033940905e-07,
+      "loss": 0.0001,
+      "step": 23802
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.078164480232074e-07,
+      "loss": 0.0021,
+      "step": 23803
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.059366861135951e-07,
+      "loss": 0.0013,
+      "step": 23804
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.040594176966452e-07,
+      "loss": 0.0019,
+      "step": 23805
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.021846428037325e-07,
+      "loss": 0.0002,
+      "step": 23806
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.003123614661487e-07,
+      "loss": 0.0001,
+      "step": 23807
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.984425737151689e-07,
+      "loss": 0.0003,
+      "step": 23808
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.965752795820179e-07,
+      "loss": 0.0027,
+      "step": 23809
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.947104790979041e-07,
+      "loss": 0.0009,
+      "step": 23810
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.928481722939527e-07,
+      "loss": 0.0006,
+      "step": 23811
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.909883592012721e-07,
+      "loss": 0.0002,
+      "step": 23812
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.891310398509375e-07,
+      "loss": 0.0001,
+      "step": 23813
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.872762142739407e-07,
+      "loss": 0.0002,
+      "step": 23814
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.854238825012904e-07,
+      "loss": 0.0016,
+      "step": 23815
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.835740445639115e-07,
+      "loss": 0.0013,
+      "step": 23816
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.817267004926962e-07,
+      "loss": 0.0005,
+      "step": 23817
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.798818503185199e-07,
+      "loss": 0.0006,
+      "step": 23818
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.780394940721578e-07,
+      "loss": 0.0003,
+      "step": 23819
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.761996317844021e-07,
+      "loss": 0.0016,
+      "step": 23820
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.743622634859946e-07,
+      "loss": 0.0042,
+      "step": 23821
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.725273892075944e-07,
+      "loss": 0.0001,
+      "step": 23822
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.706950089798768e-07,
+      "loss": 0.0004,
+      "step": 23823
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.688651228334341e-07,
+      "loss": 0.0013,
+      "step": 23824
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.670377307988084e-07,
+      "loss": 0.0011,
+      "step": 23825
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.652128329065421e-07,
+      "loss": 0.0026,
+      "step": 23826
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.633904291871272e-07,
+      "loss": 0.0005,
+      "step": 23827
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.615705196709731e-07,
+      "loss": 0.0018,
+      "step": 23828
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.597531043884885e-07,
+      "loss": 0.0008,
+      "step": 23829
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.579381833700326e-07,
+      "loss": 0.0036,
+      "step": 23830
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.561257566458978e-07,
+      "loss": 0.0003,
+      "step": 23831
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.543158242463764e-07,
+      "loss": 0.0024,
+      "step": 23832
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.525083862016945e-07,
+      "loss": 0.0028,
+      "step": 23833
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.507034425420443e-07,
+      "loss": 0.0002,
+      "step": 23834
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.489009932975519e-07,
+      "loss": 0.0007,
+      "step": 23835
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.471010384983266e-07,
+      "loss": 0.0001,
+      "step": 23836
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.453035781744443e-07,
+      "loss": 0.0021,
+      "step": 23837
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.435086123559307e-07,
+      "loss": 0.0002,
+      "step": 23838
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.417161410727456e-07,
+      "loss": 0.0041,
+      "step": 23839
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.399261643548315e-07,
+      "loss": 0.0008,
+      "step": 23840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.381386822320811e-07,
+      "loss": 0.0014,
+      "step": 23841
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.363536947343706e-07,
+      "loss": 0.0031,
+      "step": 23842
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.345712018914762e-07,
+      "loss": 0.0002,
+      "step": 23843
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.327912037332072e-07,
+      "loss": 0.0018,
+      "step": 23844
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.310137002892734e-07,
+      "loss": 0.0001,
+      "step": 23845
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.292386915893676e-07,
+      "loss": 0.0029,
+      "step": 23846
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.274661776631162e-07,
+      "loss": 0.0006,
+      "step": 23847
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.256961585401621e-07,
+      "loss": 0.0184,
+      "step": 23848
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.239286342500316e-07,
+      "loss": 0.0005,
+      "step": 23849
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.221636048222678e-07,
+      "loss": 0.0011,
+      "step": 23850
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.204010702863471e-07,
+      "loss": 0.0003,
+      "step": 23851
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.186410306716794e-07,
+      "loss": 0.0,
+      "step": 23852
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.168834860077077e-07,
+      "loss": 0.0005,
+      "step": 23853
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.151284363237585e-07,
+      "loss": 0.0047,
+      "step": 23854
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.133758816491419e-07,
+      "loss": 0.0033,
+      "step": 23855
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.116258220131343e-07,
+      "loss": 0.0017,
+      "step": 23856
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.098782574449624e-07,
+      "loss": 0.0109,
+      "step": 23857
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.081331879738194e-07,
+      "loss": 0.0011,
+      "step": 23858
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.063906136288654e-07,
+      "loss": 0.0029,
+      "step": 23859
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.046505344391772e-07,
+      "loss": 0.0005,
+      "step": 23860
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.029129504338315e-07,
+      "loss": 0.0026,
+      "step": 23861
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.011778616418384e-07,
+      "loss": 0.0021,
+      "step": 23862
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.994452680921913e-07,
+      "loss": 0.0002,
+      "step": 23863
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.977151698138172e-07,
+      "loss": 0.0054,
+      "step": 23864
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.959875668356262e-07,
+      "loss": 0.0016,
+      "step": 23865
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.94262459186462e-07,
+      "loss": 0.0003,
+      "step": 23866
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.925398468951181e-07,
+      "loss": 0.0018,
+      "step": 23867
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.908197299904049e-07,
+      "loss": 0.0003,
+      "step": 23868
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.891021085010328e-07,
+      "loss": 0.0015,
+      "step": 23869
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.873869824556787e-07,
+      "loss": 0.0004,
+      "step": 23870
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.856743518830031e-07,
+      "loss": 0.0008,
+      "step": 23871
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.839642168115999e-07,
+      "loss": 0.001,
+      "step": 23872
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.82256577270046e-07,
+      "loss": 0.0078,
+      "step": 23873
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.805514332868522e-07,
+      "loss": 0.0065,
+      "step": 23874
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.788487848904955e-07,
+      "loss": 0.0012,
+      "step": 23875
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.771486321094199e-07,
+      "loss": 0.0024,
+      "step": 23876
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.754509749720193e-07,
+      "loss": 0.0006,
+      "step": 23877
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.737558135066378e-07,
+      "loss": 0.005,
+      "step": 23878
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.720631477415861e-07,
+      "loss": 0.007,
+      "step": 23879
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.703729777051746e-07,
+      "loss": 0.0001,
+      "step": 23880
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.686853034255811e-07,
+      "loss": 0.0018,
+      "step": 23881
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.670001249310163e-07,
+      "loss": 0.0001,
+      "step": 23882
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.653174422496243e-07,
+      "loss": 0.001,
+      "step": 23883
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.636372554095159e-07,
+      "loss": 0.0005,
+      "step": 23884
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.61959564438752e-07,
+      "loss": 0.0017,
+      "step": 23885
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.602843693653436e-07,
+      "loss": 0.0005,
+      "step": 23886
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.586116702172683e-07,
+      "loss": 0.0005,
+      "step": 23887
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.569414670224703e-07,
+      "loss": 0.0002,
+      "step": 23888
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.552737598088275e-07,
+      "loss": 0.0022,
+      "step": 23889
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.536085486042341e-07,
+      "loss": 0.0005,
+      "step": 23890
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.519458334364679e-07,
+      "loss": 0.0001,
+      "step": 23891
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.502856143333068e-07,
+      "loss": 0.001,
+      "step": 23892
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.486278913224951e-07,
+      "loss": 0.0129,
+      "step": 23893
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.469726644316941e-07,
+      "loss": 0.0053,
+      "step": 23894
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.453199336885649e-07,
+      "loss": 0.0006,
+      "step": 23895
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.436696991207024e-07,
+      "loss": 0.0005,
+      "step": 23896
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.420219607557008e-07,
+      "loss": 0.0001,
+      "step": 23897
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.403767186210218e-07,
+      "loss": 0.0001,
+      "step": 23898
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.387339727441931e-07,
+      "loss": 0.0022,
+      "step": 23899
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.370937231526429e-07,
+      "loss": 0.0025,
+      "step": 23900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.354559698737493e-07,
+      "loss": 0.0035,
+      "step": 23901
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.338207129348737e-07,
+      "loss": 0.0003,
+      "step": 23902
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.321879523633277e-07,
+      "loss": 0.004,
+      "step": 23903
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.305576881863894e-07,
+      "loss": 0.0025,
+      "step": 23904
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.289299204312703e-07,
+      "loss": 0.0014,
+      "step": 23905
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.27304649125182e-07,
+      "loss": 0.0046,
+      "step": 23906
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.256818742952362e-07,
+      "loss": 0.0006,
+      "step": 23907
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.240615959685778e-07,
+      "loss": 0.001,
+      "step": 23908
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.224438141722353e-07,
+      "loss": 0.0009,
+      "step": 23909
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.20828528933237e-07,
+      "loss": 0.0005,
+      "step": 23910
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.192157402785779e-07,
+      "loss": 0.0002,
+      "step": 23911
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.176054482351699e-07,
+      "loss": 0.0016,
+      "step": 23912
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.159976528299248e-07,
+      "loss": 0.0045,
+      "step": 23913
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.143923540896877e-07,
+      "loss": 0.0033,
+      "step": 23914
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.127895520412706e-07,
+      "loss": 0.001,
+      "step": 23915
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.11189246711452e-07,
+      "loss": 0.0015,
+      "step": 23916
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.095914381269439e-07,
+      "loss": 0.0021,
+      "step": 23917
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.079961263144417e-07,
+      "loss": 0.0091,
+      "step": 23918
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.064033113005906e-07,
+      "loss": 0.0001,
+      "step": 23919
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.048129931120027e-07,
+      "loss": 0.003,
+      "step": 23920
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.032251717752233e-07,
+      "loss": 0.0013,
+      "step": 23921
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.016398473167815e-07,
+      "loss": 0.0004,
+      "step": 23922
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.00057019763156e-07,
+      "loss": 0.0022,
+      "step": 23923
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.984766891407754e-07,
+      "loss": 0.0019,
+      "step": 23924
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.968988554760356e-07,
+      "loss": 0.0002,
+      "step": 23925
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.953235187952986e-07,
+      "loss": 0.0007,
+      "step": 23926
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.937506791248435e-07,
+      "loss": 0.0007,
+      "step": 23927
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.921803364909993e-07,
+      "loss": 0.0015,
+      "step": 23928
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.906124909199449e-07,
+      "loss": 0.0001,
+      "step": 23929
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.890471424378761e-07,
+      "loss": 0.001,
+      "step": 23930
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.874842910709553e-07,
+      "loss": 0.0004,
+      "step": 23931
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.85923936845245e-07,
+      "loss": 0.0005,
+      "step": 23932
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.843660797868576e-07,
+      "loss": 0.0016,
+      "step": 23933
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.82810719921789e-07,
+      "loss": 0.0001,
+      "step": 23934
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.812578572760017e-07,
+      "loss": 0.0001,
+      "step": 23935
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.797074918754584e-07,
+      "loss": 0.0001,
+      "step": 23936
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.781596237460217e-07,
+      "loss": 0.0025,
+      "step": 23937
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.766142529135708e-07,
+      "loss": 0.0012,
+      "step": 23938
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7507137940391847e-07,
+      "loss": 0.0001,
+      "step": 23939
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.735310032428108e-07,
+      "loss": 0.0008,
+      "step": 23940
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7199312445597714e-07,
+      "loss": 0.0034,
+      "step": 23941
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.704577430691303e-07,
+      "loss": 0.0001,
+      "step": 23942
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6892485910788314e-07,
+      "loss": 0.0031,
+      "step": 23943
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.673944725978651e-07,
+      "loss": 0.0004,
+      "step": 23944
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.658665835646058e-07,
+      "loss": 0.0004,
+      "step": 23945
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.643411920336515e-07,
+      "loss": 0.0002,
+      "step": 23946
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.628182980304651e-07,
+      "loss": 0.0003,
+      "step": 23947
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6129790158047633e-07,
+      "loss": 0.0011,
+      "step": 23948
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.597800027090981e-07,
+      "loss": 0.0005,
+      "step": 23949
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.5826460144167685e-07,
+      "loss": 0.0038,
+      "step": 23950
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.567516978034924e-07,
+      "loss": 0.0006,
+      "step": 23951
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.552412918198578e-07,
+      "loss": 0.0001,
+      "step": 23952
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.5373338351596955e-07,
+      "loss": 0.002,
+      "step": 23953
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.5222797291702415e-07,
+      "loss": 0.0028,
+      "step": 23954
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.507250600481682e-07,
+      "loss": 0.0016,
+      "step": 23955
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4922464493449826e-07,
+      "loss": 0.0002,
+      "step": 23956
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4772672760107765e-07,
+      "loss": 0.0029,
+      "step": 23957
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.46231308072903e-07,
+      "loss": 0.0006,
+      "step": 23958
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.447383863749876e-07,
+      "loss": 0.0001,
+      "step": 23959
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4324796253222825e-07,
+      "loss": 0.0059,
+      "step": 23960
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4176003656955504e-07,
+      "loss": 0.0002,
+      "step": 23961
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4027460851178143e-07,
+      "loss": 0.0012,
+      "step": 23962
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.387916783837542e-07,
+      "loss": 0.0006,
+      "step": 23963
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.373112462102202e-07,
+      "loss": 0.0032,
+      "step": 23964
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.35833312015893e-07,
+      "loss": 0.0015,
+      "step": 23965
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.343578758254862e-07,
+      "loss": 0.0006,
+      "step": 23966
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.328849376636134e-07,
+      "loss": 0.0028,
+      "step": 23967
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.3141449755490497e-07,
+      "loss": 0.0088,
+      "step": 23968
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2994655552389126e-07,
+      "loss": 0.001,
+      "step": 23969
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.284811115951192e-07,
+      "loss": 0.0037,
+      "step": 23970
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2701816579303606e-07,
+      "loss": 0.0005,
+      "step": 23971
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.255577181421055e-07,
+      "loss": 0.0025,
+      "step": 23972
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2409976866669137e-07,
+      "loss": 0.0033,
+      "step": 23973
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2264431739115756e-07,
+      "loss": 0.0002,
+      "step": 23974
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.211913643398013e-07,
+      "loss": 0.0006,
+      "step": 23975
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.197409095369031e-07,
+      "loss": 0.003,
+      "step": 23976
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.182929530066936e-07,
+      "loss": 0.0007,
+      "step": 23977
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.168474947733369e-07,
+      "loss": 0.0022,
+      "step": 23978
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.154045348609969e-07,
+      "loss": 0.0031,
+      "step": 23979
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1396407329375435e-07,
+      "loss": 0.0007,
+      "step": 23980
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.125261100956734e-07,
+      "loss": 0.0008,
+      "step": 23981
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1109064529078474e-07,
+      "loss": 0.0033,
+      "step": 23982
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0965767890303593e-07,
+      "loss": 0.0034,
+      "step": 23983
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.082272109563911e-07,
+      "loss": 0.0109,
+      "step": 23984
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0679924147469787e-07,
+      "loss": 0.0046,
+      "step": 23985
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.053737704818705e-07,
+      "loss": 0.0019,
+      "step": 23986
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0395079800165653e-07,
+      "loss": 0.0033,
+      "step": 23987
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.025303240578537e-07,
+      "loss": 0.002,
+      "step": 23988
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.011123486741763e-07,
+      "loss": 0.0017,
+      "step": 23989
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.9969687187432206e-07,
+      "loss": 0.0016,
+      "step": 23990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.9828389368192217e-07,
+      "loss": 0.0025,
+      "step": 23991
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.9687341412055765e-07,
+      "loss": 0.0003,
+      "step": 23992
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.9546543321380963e-07,
+      "loss": 0.0009,
+      "step": 23993
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.940599509851927e-07,
+      "loss": 0.0017,
+      "step": 23994
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.9265696745817143e-07,
+      "loss": 0.0008,
+      "step": 23995
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.91256482656177e-07,
+      "loss": 0.0045,
+      "step": 23996
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.8985849660262415e-07,
+      "loss": 0.0019,
+      "step": 23997
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.884630093208274e-07,
+      "loss": 0.0001,
+      "step": 23998
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.870700208341182e-07,
+      "loss": 0.0001,
+      "step": 23999
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.856795311657446e-07,
+      "loss": 0.0073,
+      "step": 24000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.842915403389546e-07,
+      "loss": 0.0032,
+      "step": 24001
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.829060483768964e-07,
+      "loss": 0.0031,
+      "step": 24002
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.815230553027515e-07,
+      "loss": 0.005,
+      "step": 24003
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.8014256113958477e-07,
+      "loss": 0.0016,
+      "step": 24004
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.78764565910461e-07,
+      "loss": 0.0023,
+      "step": 24005
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.773890696383952e-07,
+      "loss": 0.0002,
+      "step": 24006
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.7601607234636897e-07,
+      "loss": 0.003,
+      "step": 24007
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.74645574057314e-07,
+      "loss": 0.0039,
+      "step": 24008
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.732775747941119e-07,
+      "loss": 0.004,
+      "step": 24009
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.719120745796278e-07,
+      "loss": 0.0009,
+      "step": 24010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.7054907343664344e-07,
+      "loss": 0.0002,
+      "step": 24011
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6918857138794055e-07,
+      "loss": 0.0005,
+      "step": 24012
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6783056845623437e-07,
+      "loss": 0.0011,
+      "step": 24013
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6647506466420673e-07,
+      "loss": 0.0032,
+      "step": 24014
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.651220600344895e-07,
+      "loss": 0.0009,
+      "step": 24015
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.63771554589698e-07,
+      "loss": 0.0001,
+      "step": 24016
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6242354835238073e-07,
+      "loss": 0.0006,
+      "step": 24017
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6107804134505315e-07,
+      "loss": 0.0,
+      "step": 24018
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5973503359018053e-07,
+      "loss": 0.0003,
+      "step": 24019
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.583945251102116e-07,
+      "loss": 0.0033,
+      "step": 24020
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5705651592751183e-07,
+      "loss": 0.0035,
+      "step": 24021
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5572100606444667e-07,
+      "loss": 0.0003,
+      "step": 24022
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5438799554329823e-07,
+      "loss": 0.0041,
+      "step": 24023
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.530574843863654e-07,
+      "loss": 0.0014,
+      "step": 24024
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.517294726158304e-07,
+      "loss": 0.0041,
+      "step": 24025
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.504039602539088e-07,
+      "loss": 0.0005,
+      "step": 24026
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4908094732273294e-07,
+      "loss": 0.0007,
+      "step": 24027
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4776043384438514e-07,
+      "loss": 0.0006,
+      "step": 24028
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4644241984091436e-07,
+      "loss": 0.0003,
+      "step": 24029
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4512690533435304e-07,
+      "loss": 0.0004,
+      "step": 24030
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4381389034666696e-07,
+      "loss": 0.0014,
+      "step": 24031
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.425033748997885e-07,
+      "loss": 0.0005,
+      "step": 24032
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4119535901560024e-07,
+      "loss": 0.0027,
+      "step": 24033
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3988984271595133e-07,
+      "loss": 0.0045,
+      "step": 24034
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3858682602264097e-07,
+      "loss": 0.0006,
+      "step": 24035
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3728630895743514e-07,
+      "loss": 0.0008,
+      "step": 24036
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3598829154206644e-07,
+      "loss": 0.0004,
+      "step": 24037
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.346927737981842e-07,
+      "loss": 0.0005,
+      "step": 24038
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.33399755747471e-07,
+      "loss": 0.0002,
+      "step": 24039
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3210923741147644e-07,
+      "loss": 0.0004,
+      "step": 24040
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3082121881178316e-07,
+      "loss": 0.0076,
+      "step": 24041
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2953569996990724e-07,
+      "loss": 0.0001,
+      "step": 24042
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.282526809072983e-07,
+      "loss": 0.0002,
+      "step": 24043
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.269721616453891e-07,
+      "loss": 0.001,
+      "step": 24044
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2569414220557923e-07,
+      "loss": 0.0003,
+      "step": 24045
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2441862260921825e-07,
+      "loss": 0.0016,
+      "step": 24046
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.231456028775725e-07,
+      "loss": 0.0011,
+      "step": 24047
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.218750830319583e-07,
+      "loss": 0.0016,
+      "step": 24048
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2060706309355865e-07,
+      "loss": 0.0032,
+      "step": 24049
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.1934154308355666e-07,
+      "loss": 0.0003,
+      "step": 24050
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.1807852302308534e-07,
+      "loss": 0.0027,
+      "step": 24051
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.168180029332612e-07,
+      "loss": 0.0001,
+      "step": 24052
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.155599828351174e-07,
+      "loss": 0.0015,
+      "step": 24053
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.1430446274967047e-07,
+      "loss": 0.0012,
+      "step": 24054
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.130514426979036e-07,
+      "loss": 0.0042,
+      "step": 24055
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.118009227007168e-07,
+      "loss": 0.0004,
+      "step": 24056
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.105529027790099e-07,
+      "loss": 0.0004,
+      "step": 24057
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.09307382953633e-07,
+      "loss": 0.0014,
+      "step": 24058
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.080643632453861e-07,
+      "loss": 0.0003,
+      "step": 24059
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.068238436750359e-07,
+      "loss": 0.0031,
+      "step": 24060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.055858242632825e-07,
+      "loss": 0.0004,
+      "step": 24061
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.043503050308094e-07,
+      "loss": 0.0047,
+      "step": 24062
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.031172859982667e-07,
+      "loss": 0.0004,
+      "step": 24063
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.018867671862213e-07,
+      "loss": 0.0004,
+      "step": 24064
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.006587486152734e-07,
+      "loss": 0.0003,
+      "step": 24065
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.994332303058733e-07,
+      "loss": 0.0018,
+      "step": 24066
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.9821021227853794e-07,
+      "loss": 0.0001,
+      "step": 24067
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.969896945536676e-07,
+      "loss": 0.0084,
+      "step": 24068
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.9577167715164606e-07,
+      "loss": 0.0055,
+      "step": 24069
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.945561600928237e-07,
+      "loss": 0.0011,
+      "step": 24070
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.933431433975175e-07,
+      "loss": 0.0002,
+      "step": 24071
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.921326270859614e-07,
+      "loss": 0.0001,
+      "step": 24072
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.909246111783725e-07,
+      "loss": 0.0019,
+      "step": 24073
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.89719095694968e-07,
+      "loss": 0.0001,
+      "step": 24074
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8851608065583174e-07,
+      "loss": 0.0011,
+      "step": 24075
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.873155660810811e-07,
+      "loss": 0.0013,
+      "step": 24076
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.861175519907499e-07,
+      "loss": 0.0018,
+      "step": 24077
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.849220384048889e-07,
+      "loss": 0.0018,
+      "step": 24078
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8372902534341545e-07,
+      "loss": 0.0002,
+      "step": 24079
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8253851282628026e-07,
+      "loss": 0.0007,
+      "step": 24080
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.813505008733674e-07,
+      "loss": 0.0033,
+      "step": 24081
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8016498950451107e-07,
+      "loss": 0.0027,
+      "step": 24082
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7898197873952864e-07,
+      "loss": 0.0002,
+      "step": 24083
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7780146859815444e-07,
+      "loss": 0.0011,
+      "step": 24084
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7662345910012263e-07,
+      "loss": 0.0045,
+      "step": 24085
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.754479502651008e-07,
+      "loss": 0.0022,
+      "step": 24086
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7427494211272326e-07,
+      "loss": 0.004,
+      "step": 24087
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.731044346625744e-07,
+      "loss": 0.0009,
+      "step": 24088
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7193642793422175e-07,
+      "loss": 0.0007,
+      "step": 24089
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.707709219471499e-07,
+      "loss": 0.0043,
+      "step": 24090
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6960791672084314e-07,
+      "loss": 0.0021,
+      "step": 24091
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6844741227471934e-07,
+      "loss": 0.0026,
+      "step": 24092
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6728940862816295e-07,
+      "loss": 0.0003,
+      "step": 24093
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6613390580052517e-07,
+      "loss": 0.0009,
+      "step": 24094
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6498090381107394e-07,
+      "loss": 0.0003,
+      "step": 24095
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.638304026790938e-07,
+      "loss": 0.0003,
+      "step": 24096
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6268240242380277e-07,
+      "loss": 0.0018,
+      "step": 24097
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6153690306435215e-07,
+      "loss": 0.001,
+      "step": 24098
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6039390461989327e-07,
+      "loss": 0.0061,
+      "step": 24099
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5925340710951095e-07,
+      "loss": 0.0001,
+      "step": 24100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.581154105522398e-07,
+      "loss": 0.0036,
+      "step": 24101
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5697991496711476e-07,
+      "loss": 0.0027,
+      "step": 24102
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.558469203730873e-07,
+      "loss": 0.0026,
+      "step": 24103
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547164267890589e-07,
+      "loss": 0.004,
+      "step": 24104
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.535884342339478e-07,
+      "loss": 0.0098,
+      "step": 24105
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.524629427265723e-07,
+      "loss": 0.0005,
+      "step": 24106
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5133995228573403e-07,
+      "loss": 0.0006,
+      "step": 24107
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5021946293018456e-07,
+      "loss": 0.0039,
+      "step": 24108
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.491014746786424e-07,
+      "loss": 0.0,
+      "step": 24109
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4798598754977583e-07,
+      "loss": 0.0004,
+      "step": 24110
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4687300156223667e-07,
+      "loss": 0.0001,
+      "step": 24111
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4576251673459337e-07,
+      "loss": 0.0009,
+      "step": 24112
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4465453308538107e-07,
+      "loss": 0.0004,
+      "step": 24113
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4354905063313503e-07,
+      "loss": 0.0094,
+      "step": 24114
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4244606939630706e-07,
+      "loss": 0.0001,
+      "step": 24115
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4134558939331585e-07,
+      "loss": 0.0003,
+      "step": 24116
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.402476106425466e-07,
+      "loss": 0.0026,
+      "step": 24117
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3915213316231806e-07,
+      "loss": 0.0011,
+      "step": 24118
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.380591569709489e-07,
+      "loss": 0.0002,
+      "step": 24119
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3696868208669117e-07,
+      "loss": 0.0021,
+      "step": 24120
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3588070852776364e-07,
+      "loss": 0.0006,
+      "step": 24121
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3479523631231845e-07,
+      "loss": 0.0025,
+      "step": 24122
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3371226545849108e-07,
+      "loss": 0.0001,
+      "step": 24123
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.326317959843671e-07,
+      "loss": 0.0015,
+      "step": 24124
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3155382790801537e-07,
+      "loss": 0.0014,
+      "step": 24125
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3047836124740482e-07,
+      "loss": 0.0003,
+      "step": 24126
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2940539602053776e-07,
+      "loss": 0.0069,
+      "step": 24127
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.283349322452832e-07,
+      "loss": 0.0037,
+      "step": 24128
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2726696993957682e-07,
+      "loss": 0.0083,
+      "step": 24129
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2620150912122102e-07,
+      "loss": 0.0025,
+      "step": 24130
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.251385498080016e-07,
+      "loss": 0.0008,
+      "step": 24131
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2407809201770433e-07,
+      "loss": 0.0006,
+      "step": 24132
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2302013576801502e-07,
+      "loss": 0.0017,
+      "step": 24133
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2196468107663623e-07,
+      "loss": 0.0007,
+      "step": 24134
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2091172796115386e-07,
+      "loss": 0.0002,
+      "step": 24135
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.198612764391705e-07,
+      "loss": 0.0004,
+      "step": 24136
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.188133265282388e-07,
+      "loss": 0.0004,
+      "step": 24137
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.177678782458614e-07,
+      "loss": 0.0021,
+      "step": 24138
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.16724931609491e-07,
+      "loss": 0.0054,
+      "step": 24139
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1568448663654704e-07,
+      "loss": 0.0234,
+      "step": 24140
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1464654334441555e-07,
+      "loss": 0.0067,
+      "step": 24141
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1361110175043272e-07,
+      "loss": 0.0014,
+      "step": 24142
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1257816187186805e-07,
+      "loss": 0.0015,
+      "step": 24143
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.115477237260077e-07,
+      "loss": 0.0002,
+      "step": 24144
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1051978733003795e-07,
+      "loss": 0.0006,
+      "step": 24145
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.09494352701145e-07,
+      "loss": 0.0001,
+      "step": 24146
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.084714198564319e-07,
+      "loss": 0.0001,
+      "step": 24147
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0745098881301825e-07,
+      "loss": 0.0003,
+      "step": 24148
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0643305958790713e-07,
+      "loss": 0.0008,
+      "step": 24149
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0541763219811824e-07,
+      "loss": 0.0005,
+      "step": 24150
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0440470666062135e-07,
+      "loss": 0.0006,
+      "step": 24151
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0339428299231963e-07,
+      "loss": 0.0006,
+      "step": 24152
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0238636121009954e-07,
+      "loss": 0.0003,
+      "step": 24153
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0138094133079762e-07,
+      "loss": 0.0012,
+      "step": 24154
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0037802337118381e-07,
+      "loss": 0.0001,
+      "step": 24155
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.99377607348028e-07,
+      "loss": 0.0006,
+      "step": 24156
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.983796932780335e-07,
+      "loss": 0.0007,
+      "step": 24157
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.973842811778703e-07,
+      "loss": 0.0001,
+      "step": 24158
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9639137106417513e-07,
+      "loss": 0.0017,
+      "step": 24159
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9540096295350137e-07,
+      "loss": 0.005,
+      "step": 24160
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9441305686241914e-07,
+      "loss": 0.0005,
+      "step": 24161
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.934276528073986e-07,
+      "loss": 0.0013,
+      "step": 24162
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9244475080494316e-07,
+      "loss": 0.001,
+      "step": 24163
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.914643508714231e-07,
+      "loss": 0.0047,
+      "step": 24164
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9048645302324195e-07,
+      "loss": 0.0023,
+      "step": 24165
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8951105727670335e-07,
+      "loss": 0.0009,
+      "step": 24166
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8853816364814424e-07,
+      "loss": 0.0012,
+      "step": 24167
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.875677721537683e-07,
+      "loss": 0.0004,
+      "step": 24168
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8659988280981253e-07,
+      "loss": 0.0015,
+      "step": 24169
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8563449563241407e-07,
+      "loss": 0.0164,
+      "step": 24170
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8467161063772662e-07,
+      "loss": 0.0024,
+      "step": 24171
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8371122784182068e-07,
+      "loss": 0.0028,
+      "step": 24172
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8275334726073343e-07,
+      "loss": 0.0024,
+      "step": 24173
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8179796891046871e-07,
+      "loss": 0.0008,
+      "step": 24174
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8084509280698046e-07,
+      "loss": 0.0001,
+      "step": 24175
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7989471896618922e-07,
+      "loss": 0.0024,
+      "step": 24176
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7894684740394905e-07,
+      "loss": 0.0075,
+      "step": 24177
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7800147813609722e-07,
+      "loss": 0.0016,
+      "step": 24178
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7705861117843778e-07,
+      "loss": 0.0009,
+      "step": 24179
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7611824654672478e-07,
+      "loss": 0.0014,
+      "step": 24180
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7518038425662907e-07,
+      "loss": 0.0002,
+      "step": 24181
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7424502432383802e-07,
+      "loss": 0.0009,
+      "step": 24182
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7331216676398918e-07,
+      "loss": 0.002,
+      "step": 24183
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.723818115926201e-07,
+      "loss": 0.0074,
+      "step": 24184
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.714539588253183e-07,
+      "loss": 0.0056,
+      "step": 24185
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7052860847753813e-07,
+      "loss": 0.0001,
+      "step": 24186
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6960576056476715e-07,
+      "loss": 0.0002,
+      "step": 24187
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6868541510239308e-07,
+      "loss": 0.0014,
+      "step": 24188
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.677675721058036e-07,
+      "loss": 0.0007,
+      "step": 24189
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.668522315903198e-07,
+      "loss": 0.0005,
+      "step": 24190
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6593939357124607e-07,
+      "loss": 0.0017,
+      "step": 24191
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6502905806380362e-07,
+      "loss": 0.0004,
+      "step": 24192
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6412122508321356e-07,
+      "loss": 0.0062,
+      "step": 24193
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6321589464464714e-07,
+      "loss": 0.0008,
+      "step": 24194
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6231306676319223e-07,
+      "loss": 0.0006,
+      "step": 24195
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6141274145395344e-07,
+      "loss": 0.0027,
+      "step": 24196
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6051491873196876e-07,
+      "loss": 0.0039,
+      "step": 24197
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5961959861222617e-07,
+      "loss": 0.0012,
+      "step": 24198
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.587267811096804e-07,
+      "loss": 0.0005,
+      "step": 24199
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5783646623923618e-07,
+      "loss": 0.0013,
+      "step": 24200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5694865401576495e-07,
+      "loss": 0.001,
+      "step": 24201
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5606334445412148e-07,
+      "loss": 0.0017,
+      "step": 24202
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5518053756906067e-07,
+      "loss": 0.0009,
+      "step": 24203
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5430023337532072e-07,
+      "loss": 0.0026,
+      "step": 24204
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.534224318876398e-07,
+      "loss": 0.0001,
+      "step": 24205
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5254713312065625e-07,
+      "loss": 0.0002,
+      "step": 24206
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5167433708899168e-07,
+      "loss": 0.0025,
+      "step": 24207
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5080404380721777e-07,
+      "loss": 0.0008,
+      "step": 24208
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4993625328988955e-07,
+      "loss": 0.0001,
+      "step": 24209
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.490709655514788e-07,
+      "loss": 0.0077,
+      "step": 24210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4820818060644056e-07,
+      "loss": 0.0057,
+      "step": 24211
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.473478984691967e-07,
+      "loss": 0.0029,
+      "step": 24212
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.46490119154119e-07,
+      "loss": 0.002,
+      "step": 24213
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4563484267551272e-07,
+      "loss": 0.0002,
+      "step": 24214
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4478206904768308e-07,
+      "loss": 0.0001,
+      "step": 24215
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4393179828486867e-07,
+      "loss": 0.0002,
+      "step": 24216
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.430840304012415e-07,
+      "loss": 0.0002,
+      "step": 24217
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4223876541100688e-07,
+      "loss": 0.003,
+      "step": 24218
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4139600332827018e-07,
+      "loss": 0.0006,
+      "step": 24219
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.405557441670868e-07,
+      "loss": 0.0001,
+      "step": 24220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3971798794149558e-07,
+      "loss": 0.0004,
+      "step": 24221
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3888273466550193e-07,
+      "loss": 0.0028,
+      "step": 24222
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.380499843530447e-07,
+      "loss": 0.0011,
+      "step": 24223
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3721973701804612e-07,
+      "loss": 0.0002,
+      "step": 24224
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3639199267436173e-07,
+      "loss": 0.0009,
+      "step": 24225
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3556675133581384e-07,
+      "loss": 0.0001,
+      "step": 24226
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3474401301619143e-07,
+      "loss": 0.0017,
+      "step": 24227
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3392377772923347e-07,
+      "loss": 0.0007,
+      "step": 24228
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3310604548864566e-07,
+      "loss": 0.0003,
+      "step": 24229
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.322908163080838e-07,
+      "loss": 0.0014,
+      "step": 24230
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3147809020115364e-07,
+      "loss": 0.0019,
+      "step": 24231
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.306678671814443e-07,
+      "loss": 0.0003,
+      "step": 24232
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2986014726246163e-07,
+      "loss": 0.0009,
+      "step": 24233
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2905493045772818e-07,
+      "loss": 0.0015,
+      "step": 24234
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.282522167806832e-07,
+      "loss": 0.0001,
+      "step": 24235
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2745200624471596e-07,
+      "loss": 0.0029,
+      "step": 24236
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2665429886321576e-07,
+      "loss": 0.002,
+      "step": 24237
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2585909464948862e-07,
+      "loss": 0.0065,
+      "step": 24238
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.250663936168239e-07,
+      "loss": 0.0009,
+      "step": 24239
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2427619577846104e-07,
+      "loss": 0.0049,
+      "step": 24240
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2348850114760612e-07,
+      "loss": 0.0012,
+      "step": 24241
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2270330973738196e-07,
+      "loss": 0.0008,
+      "step": 24242
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2192062156094474e-07,
+      "loss": 0.0017,
+      "step": 24243
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.21140436631334e-07,
+      "loss": 0.0029,
+      "step": 24244
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2036275496160597e-07,
+      "loss": 0.0057,
+      "step": 24245
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1958757656473362e-07,
+      "loss": 0.0036,
+      "step": 24246
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1881490145367322e-07,
+      "loss": 0.0001,
+      "step": 24247
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1804472964131451e-07,
+      "loss": 0.0016,
+      "step": 24248
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1727706114053048e-07,
+      "loss": 0.0038,
+      "step": 24249
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1651189596416088e-07,
+      "loss": 0.0001,
+      "step": 24250
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1574923412494553e-07,
+      "loss": 0.0016,
+      "step": 24251
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1498907563565751e-07,
+      "loss": 0.0018,
+      "step": 24252
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.142314205089867e-07,
+      "loss": 0.001,
+      "step": 24253
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1347626875755633e-07,
+      "loss": 0.0003,
+      "step": 24254
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1272362039402294e-07,
+      "loss": 0.0001,
+      "step": 24255
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.119734754309265e-07,
+      "loss": 0.0067,
+      "step": 24256
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1122583388082363e-07,
+      "loss": 0.0003,
+      "step": 24257
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1048069575617102e-07,
+      "loss": 0.0006,
+      "step": 24258
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.097380610694254e-07,
+      "loss": 0.0002,
+      "step": 24259
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.089979298329935e-07,
+      "loss": 0.0009,
+      "step": 24260
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0826030205924874e-07,
+      "loss": 0.0019,
+      "step": 24261
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.075251777604813e-07,
+      "loss": 0.0004,
+      "step": 24262
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0679255694899802e-07,
+      "loss": 0.0025,
+      "step": 24263
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0606243963700579e-07,
+      "loss": 0.0001,
+      "step": 24264
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0533482583672814e-07,
+      "loss": 0.0025,
+      "step": 24265
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0460971556030539e-07,
+      "loss": 0.0028,
+      "step": 24266
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0388710881984452e-07,
+      "loss": 0.0002,
+      "step": 24267
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.031670056274192e-07,
+      "loss": 0.0029,
+      "step": 24268
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0244940599505314e-07,
+      "loss": 0.0032,
+      "step": 24269
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0173430993473675e-07,
+      "loss": 0.0058,
+      "step": 24270
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0102171745841048e-07,
+      "loss": 0.0043,
+      "step": 24271
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0031162857796482e-07,
+      "loss": 0.0017,
+      "step": 24272
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.960404330529026e-08,
+      "loss": 0.0027,
+      "step": 24273
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.889896165217737e-08,
+      "loss": 0.0028,
+      "step": 24274
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.819638363040005e-08,
+      "loss": 0.0011,
+      "step": 24275
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.749630925171558e-08,
+      "loss": 0.0015,
+      "step": 24276
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.679873852779796e-08,
+      "loss": 0.0,
+      "step": 24277
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.61036714703045e-08,
+      "loss": 0.0027,
+      "step": 24278
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.541110809084263e-08,
+      "loss": 0.0002,
+      "step": 24279
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.472104840098637e-08,
+      "loss": 0.0034,
+      "step": 24280
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.403349241224322e-08,
+      "loss": 0.0,
+      "step": 24281
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.334844013612063e-08,
+      "loss": 0.0016,
+      "step": 24282
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.26658915840428e-08,
+      "loss": 0.0001,
+      "step": 24283
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.198584676741727e-08,
+      "loss": 0.0001,
+      "step": 24284
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.130830569758496e-08,
+      "loss": 0.0005,
+      "step": 24285
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.06332683858868e-08,
+      "loss": 0.0039,
+      "step": 24286
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.996073484359711e-08,
+      "loss": 0.0083,
+      "step": 24287
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.92907050819236e-08,
+      "loss": 0.0032,
+      "step": 24288
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.862317911207395e-08,
+      "loss": 0.0024,
+      "step": 24289
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.795815694520591e-08,
+      "loss": 0.0002,
+      "step": 24290
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.729563859241062e-08,
+      "loss": 0.0035,
+      "step": 24291
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.66356240647459e-08,
+      "loss": 0.0002,
+      "step": 24292
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.597811337326954e-08,
+      "loss": 0.0027,
+      "step": 24293
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.532310652892282e-08,
+      "loss": 0.0006,
+      "step": 24294
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.467060354268029e-08,
+      "loss": 0.0009,
+      "step": 24295
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.402060442541658e-08,
+      "loss": 0.0004,
+      "step": 24296
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.337310918798967e-08,
+      "loss": 0.0001,
+      "step": 24297
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.272811784122425e-08,
+      "loss": 0.0018,
+      "step": 24298
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.208563039587834e-08,
+      "loss": 0.0018,
+      "step": 24299
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.144564686271004e-08,
+      "loss": 0.0049,
+      "step": 24300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.080816725237749e-08,
+      "loss": 0.0003,
+      "step": 24301
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.017319157555546e-08,
+      "loss": 0.0015,
+      "step": 24302
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.954071984281885e-08,
+      "loss": 0.0002,
+      "step": 24303
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.891075206475916e-08,
+      "loss": 0.0018,
+      "step": 24304
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.828328825186803e-08,
+      "loss": 0.0045,
+      "step": 24305
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.76583284146537e-08,
+      "loss": 0.0023,
+      "step": 24306
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.703587256355781e-08,
+      "loss": 0.0029,
+      "step": 24307
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.641592070893876e-08,
+      "loss": 0.0005,
+      "step": 24308
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.579847286118824e-08,
+      "loss": 0.0029,
+      "step": 24309
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.518352903061465e-08,
+      "loss": 0.0012,
+      "step": 24310
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.45710892274598e-08,
+      "loss": 0.007,
+      "step": 24311
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.396115346198217e-08,
+      "loss": 0.0008,
+      "step": 24312
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.335372174435694e-08,
+      "loss": 0.001,
+      "step": 24313
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.2748794084726e-08,
+      "loss": 0.0002,
+      "step": 24314
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.214637049321459e-08,
+      "loss": 0.0004,
+      "step": 24315
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.154645097984802e-08,
+      "loss": 0.0026,
+      "step": 24316
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.094903555466824e-08,
+      "loss": 0.0005,
+      "step": 24317
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.035412422766728e-08,
+      "loss": 0.0,
+      "step": 24318
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.976171700875388e-08,
+      "loss": 0.0009,
+      "step": 24319
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.917181390783677e-08,
+      "loss": 0.0021,
+      "step": 24320
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.858441493475808e-08,
+      "loss": 0.0091,
+      "step": 24321
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.799952009932664e-08,
+      "loss": 0.0003,
+      "step": 24322
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.74171294113346e-08,
+      "loss": 0.0001,
+      "step": 24323
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.683724288049086e-08,
+      "loss": 0.0003,
+      "step": 24324
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.625986051648768e-08,
+      "loss": 0.001,
+      "step": 24325
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.568498232896736e-08,
+      "loss": 0.0015,
+      "step": 24326
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.511260832752219e-08,
+      "loss": 0.0004,
+      "step": 24327
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.454273852171122e-08,
+      "loss": 0.0077,
+      "step": 24328
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.39753729210768e-08,
+      "loss": 0.0058,
+      "step": 24329
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.341051153506138e-08,
+      "loss": 0.003,
+      "step": 24330
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.284815437312407e-08,
+      "loss": 0.002,
+      "step": 24331
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.228830144465735e-08,
+      "loss": 0.0036,
+      "step": 24332
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.17309527589871e-08,
+      "loss": 0.0002,
+      "step": 24333
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.117610832543918e-08,
+      "loss": 0.0067,
+      "step": 24334
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.06237681532895e-08,
+      "loss": 0.0001,
+      "step": 24335
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 6.007393225176404e-08,
+      "loss": 0.0008,
+      "step": 24336
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.9526600630022126e-08,
+      "loss": 0.0024,
+      "step": 24337
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.898177329720643e-08,
+      "loss": 0.0032,
+      "step": 24338
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.843945026244301e-08,
+      "loss": 0.0025,
+      "step": 24339
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.7899631534774616e-08,
+      "loss": 0.0007,
+      "step": 24340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.736231712321071e-08,
+      "loss": 0.0006,
+      "step": 24341
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.68275070367441e-08,
+      "loss": 0.0001,
+      "step": 24342
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.629520128430098e-08,
+      "loss": 0.0029,
+      "step": 24343
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.576539987475759e-08,
+      "loss": 0.0011,
+      "step": 24344
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.523810281695684e-08,
+      "loss": 0.0007,
+      "step": 24345
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4713310119741674e-08,
+      "loss": 0.0001,
+      "step": 24346
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4191021791855086e-08,
+      "loss": 0.0001,
+      "step": 24347
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.367123784200678e-08,
+      "loss": 0.0006,
+      "step": 24348
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.315395827890645e-08,
+      "loss": 0.0004,
+      "step": 24349
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.2639183111163884e-08,
+      "loss": 0.0007,
+      "step": 24350
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.2126912347405513e-08,
+      "loss": 0.0004,
+      "step": 24351
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.16171459961745e-08,
+      "loss": 0.0027,
+      "step": 24352
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.110988406598071e-08,
+      "loss": 0.0103,
+      "step": 24353
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.060512656530069e-08,
+      "loss": 0.0051,
+      "step": 24354
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.0102873502577686e-08,
+      "loss": 0.0021,
+      "step": 24355
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9603124886171684e-08,
+      "loss": 0.0003,
+      "step": 24356
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.910588072444266e-08,
+      "loss": 0.0032,
+      "step": 24357
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.861114102571728e-08,
+      "loss": 0.0017,
+      "step": 24358
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.8118905798222305e-08,
+      "loss": 0.0002,
+      "step": 24359
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.7629175050201143e-08,
+      "loss": 0.0004,
+      "step": 24360
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.714194878981392e-08,
+      "loss": 0.0055,
+      "step": 24361
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.665722702523744e-08,
+      "loss": 0.0002,
+      "step": 24362
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.617500976451527e-08,
+      "loss": 0.001,
+      "step": 24363
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.5695297015757584e-08,
+      "loss": 0.0011,
+      "step": 24364
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.521808878692468e-08,
+      "loss": 0.0018,
+      "step": 24365
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.4743385086010167e-08,
+      "loss": 0.0017,
+      "step": 24366
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.4271185920957684e-08,
+      "loss": 0.0027,
+      "step": 24367
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.3801491299627625e-08,
+      "loss": 0.0003,
+      "step": 24368
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.333430122988035e-08,
+      "loss": 0.0008,
+      "step": 24369
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.286961571950964e-08,
+      "loss": 0.0015,
+      "step": 24370
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.240743477627595e-08,
+      "loss": 0.0003,
+      "step": 24371
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1947758407906426e-08,
+      "loss": 0.0005,
+      "step": 24372
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.149058662207827e-08,
+      "loss": 0.0035,
+      "step": 24373
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.103591942641871e-08,
+      "loss": 0.0044,
+      "step": 24374
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.058375682852166e-08,
+      "loss": 0.0001,
+      "step": 24375
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.013409883596441e-08,
+      "loss": 0.0006,
+      "step": 24376
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.968694545622431e-08,
+      "loss": 0.0002,
+      "step": 24377
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.9242296696778696e-08,
+      "loss": 0.0061,
+      "step": 24378
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8800152565054977e-08,
+      "loss": 0.0003,
+      "step": 24379
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.836051306844723e-08,
+      "loss": 0.0054,
+      "step": 24380
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7923378214299583e-08,
+      "loss": 0.0009,
+      "step": 24381
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7488748009889545e-08,
+      "loss": 0.0021,
+      "step": 24382
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.705662246251129e-08,
+      "loss": 0.0002,
+      "step": 24383
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.6627001579342397e-08,
+      "loss": 0.0018,
+      "step": 24384
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.619988536759377e-08,
+      "loss": 0.0024,
+      "step": 24385
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.5775273834376394e-08,
+      "loss": 0.0072,
+      "step": 24386
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.535316698678459e-08,
+      "loss": 0.0,
+      "step": 24387
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.493356483187937e-08,
+      "loss": 0.0014,
+      "step": 24388
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.4516467376655146e-08,
+      "loss": 0.0001,
+      "step": 24389
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.410187462810632e-08,
+      "loss": 0.0036,
+      "step": 24390
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3689786593110724e-08,
+      "loss": 0.0023,
+      "step": 24391
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.328020327859615e-08,
+      "loss": 0.0012,
+      "step": 24392
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.287312469137382e-08,
+      "loss": 0.0002,
+      "step": 24393
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.2468550838254946e-08,
+      "loss": 0.0002,
+      "step": 24394
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.2066481726000795e-08,
+      "loss": 0.0052,
+      "step": 24395
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.1666917361322654e-08,
+      "loss": 0.0002,
+      "step": 24396
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.126985775088187e-08,
+      "loss": 0.006,
+      "step": 24397
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.087530290132312e-08,
+      "loss": 0.0014,
+      "step": 24398
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.048325281924113e-08,
+      "loss": 0.0011,
+      "step": 24399
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.009370751118068e-08,
+      "loss": 0.0017,
+      "step": 24400
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.9706666983619897e-08,
+      "loss": 0.0003,
+      "step": 24401
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.9322131243070257e-08,
+      "loss": 0.0124,
+      "step": 24402
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.894010029592664e-08,
+      "loss": 0.0,
+      "step": 24403
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.8560574148567273e-08,
+      "loss": 0.0007,
+      "step": 24404
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.8183552807337085e-08,
+      "loss": 0.0001,
+      "step": 24405
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7809036278531038e-08,
+      "loss": 0.0044,
+      "step": 24406
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7437024568410794e-08,
+      "loss": 0.0001,
+      "step": 24407
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7067517683188045e-08,
+      "loss": 0.0002,
+      "step": 24408
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.6700515629024533e-08,
+      "loss": 0.0001,
+      "step": 24409
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.633601841206534e-08,
+      "loss": 0.0008,
+      "step": 24410
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.597402603838894e-08,
+      "loss": 0.0012,
+      "step": 24411
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5614538514040496e-08,
+      "loss": 0.0038,
+      "step": 24412
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5257555845031862e-08,
+      "loss": 0.002,
+      "step": 24413
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4903078037308287e-08,
+      "loss": 0.003,
+      "step": 24414
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4551105096798362e-08,
+      "loss": 0.0024,
+      "step": 24415
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.420163702939737e-08,
+      "loss": 0.0016,
+      "step": 24416
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3854673840917326e-08,
+      "loss": 0.0042,
+      "step": 24417
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3510215537170252e-08,
+      "loss": 0.0002,
+      "step": 24418
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3168262123901548e-08,
+      "loss": 0.0004,
+      "step": 24419
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.282881360683997e-08,
+      "loss": 0.0023,
+      "step": 24420
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2491869991614343e-08,
+      "loss": 0.0003,
+      "step": 24421
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2157431283886808e-08,
+      "loss": 0.0042,
+      "step": 24422
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1825497489236233e-08,
+      "loss": 0.0001,
+      "step": 24423
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1496068613208183e-08,
+      "loss": 0.0024,
+      "step": 24424
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1169144661281612e-08,
+      "loss": 0.0007,
+      "step": 24425
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.084472563895212e-08,
+      "loss": 0.0054,
+      "step": 24426
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.052281155161539e-08,
+      "loss": 0.0001,
+      "step": 24427
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.020340240465046e-08,
+      "loss": 0.0011,
+      "step": 24428
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9886498203403044e-08,
+      "loss": 0.0032,
+      "step": 24429
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.957209895315226e-08,
+      "loss": 0.0001,
+      "step": 24430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.926020465916056e-08,
+      "loss": 0.003,
+      "step": 24431
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8950815326640445e-08,
+      "loss": 0.0037,
+      "step": 24432
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8643930960737795e-08,
+      "loss": 0.0004,
+      "step": 24433
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8339551566598497e-08,
+      "loss": 0.0013,
+      "step": 24434
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8037677149285168e-08,
+      "loss": 0.0005,
+      "step": 24435
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7738307713877077e-08,
+      "loss": 0.0014,
+      "step": 24436
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7441443265336918e-08,
+      "loss": 0.0021,
+      "step": 24437
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7147083808627393e-08,
+      "loss": 0.0003,
+      "step": 24438
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6855229348694544e-08,
+      "loss": 0.0001,
+      "step": 24439
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6565879890384493e-08,
+      "loss": 0.0003,
+      "step": 24440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6279035438526712e-08,
+      "loss": 0.0021,
+      "step": 24441
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5994695997934015e-08,
+      "loss": 0.0015,
+      "step": 24442
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5712861573335954e-08,
+      "loss": 0.0069,
+      "step": 24443
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.543353216946208e-08,
+      "loss": 0.0051,
+      "step": 24444
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.515670779095868e-08,
+      "loss": 0.0008,
+      "step": 24445
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4882388442455373e-08,
+      "loss": 0.0002,
+      "step": 24446
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4610574128531837e-08,
+      "loss": 0.0061,
+      "step": 24447
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4341264853734436e-08,
+      "loss": 0.0033,
+      "step": 24448
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4074460622542915e-08,
+      "loss": 0.005,
+      "step": 24449
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3810161439437028e-08,
+      "loss": 0.0012,
+      "step": 24450
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3548367308813257e-08,
+      "loss": 0.0013,
+      "step": 24451
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3289078235068085e-08,
+      "loss": 0.0018,
+      "step": 24452
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3032294222498074e-08,
+      "loss": 0.0025,
+      "step": 24453
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2778015275416442e-08,
+      "loss": 0.001,
+      "step": 24454
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2526241398053138e-08,
+      "loss": 0.0018,
+      "step": 24455
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2276972594638112e-08,
+      "loss": 0.0005,
+      "step": 24456
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2030208869301395e-08,
+      "loss": 0.0065,
+      "step": 24457
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1785950226189666e-08,
+      "loss": 0.0005,
+      "step": 24458
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1544196669382999e-08,
+      "loss": 0.0037,
+      "step": 24459
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1304948202911501e-08,
+      "loss": 0.0046,
+      "step": 24460
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1068204830771977e-08,
+      "loss": 0.0023,
+      "step": 24461
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0833966556911266e-08,
+      "loss": 0.0005,
+      "step": 24462
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0602233385242909e-08,
+      "loss": 0.0001,
+      "step": 24463
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0373005319647132e-08,
+      "loss": 0.0015,
+      "step": 24464
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0146282363954206e-08,
+      "loss": 0.004,
+      "step": 24465
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.92206452194444e-09,
+      "loss": 0.0001,
+      "step": 24466
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.700351797348182e-09,
+      "loss": 0.0014,
+      "step": 24467
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.481144193895785e-09,
+      "loss": 0.0001,
+      "step": 24468
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.264441715234327e-09,
+      "loss": 0.003,
+      "step": 24469
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.050244364977588e-09,
+      "loss": 0.001,
+      "step": 24470
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.838552146706035e-09,
+      "loss": 0.0006,
+      "step": 24471
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.629365063966831e-09,
+      "loss": 0.0001,
+      "step": 24472
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.422683120257179e-09,
+      "loss": 0.0011,
+      "step": 24473
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.218506318991013e-09,
+      "loss": 0.0004,
+      "step": 24474
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.016834663632232e-09,
+      "loss": 0.0002,
+      "step": 24475
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.817668157511503e-09,
+      "loss": 0.001,
+      "step": 24476
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.621006803959495e-09,
+      "loss": 0.0004,
+      "step": 24477
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.4268506062735715e-09,
+      "loss": 0.0004,
+      "step": 24478
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.2351995676844795e-09,
+      "loss": 0.0006,
+      "step": 24479
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.046053691406317e-09,
+      "loss": 0.0001,
+      "step": 24480
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.8594129805865646e-09,
+      "loss": 0.0011,
+      "step": 24481
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.6752774383560525e-09,
+      "loss": 0.0021,
+      "step": 24482
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.493647067778995e-09,
+      "loss": 0.0024,
+      "step": 24483
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.314521871886302e-09,
+      "loss": 0.0039,
+      "step": 24484
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.137901853658922e-09,
+      "loss": 0.0072,
+      "step": 24485
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.963787016077803e-09,
+      "loss": 0.0003,
+      "step": 24486
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.792177362040629e-09,
+      "loss": 0.0001,
+      "step": 24487
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.6230728943951195e-09,
+      "loss": 0.0004,
+      "step": 24488
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.456473615988999e-09,
+      "loss": 0.0063,
+      "step": 24489
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.292379529586721e-09,
+      "loss": 0.0003,
+      "step": 24490
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.130790637952742e-09,
+      "loss": 0.0025,
+      "step": 24491
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.971706943751597e-09,
+      "loss": 0.0025,
+      "step": 24492
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.815128449664474e-09,
+      "loss": 0.0003,
+      "step": 24493
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.6610551583059486e-09,
+      "loss": 0.0024,
+      "step": 24494
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.5094870722406365e-09,
+      "loss": 0.0007,
+      "step": 24495
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3604241939998455e-09,
+      "loss": 0.0003,
+      "step": 24496
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.213866526081577e-09,
+      "loss": 0.0034,
+      "step": 24497
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.069814070917221e-09,
+      "loss": 0.0014,
+      "step": 24498
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.928266830938165e-09,
+      "loss": 0.0001,
+      "step": 24499
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.78922480849253e-09,
+      "loss": 0.0058,
+      "step": 24500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.6526880058951325e-09,
+      "loss": 0.0022,
+      "step": 24501
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.518656425444133e-09,
+      "loss": 0.0011,
+      "step": 24502
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3871300693710803e-09,
+      "loss": 0.0003,
+      "step": 24503
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2581089398742155e-09,
+      "loss": 0.0013,
+      "step": 24504
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.1315930391018207e-09,
+      "loss": 0.0002,
+      "step": 24505
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.0075823691688705e-09,
+      "loss": 0.0005,
+      "step": 24506
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.8860769321403797e-09,
+      "loss": 0.0014,
+      "step": 24507
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.76707673006471e-09,
+      "loss": 0.0002,
+      "step": 24508
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6505817649236095e-09,
+      "loss": 0.0001,
+      "step": 24509
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.536592038648866e-09,
+      "loss": 0.0058,
+      "step": 24510
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.425107553155614e-09,
+      "loss": 0.001,
+      "step": 24511
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.316128310309029e-09,
+      "loss": 0.0006,
+      "step": 24512
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.209654311924325e-09,
+      "loss": 0.0002,
+      "step": 24513
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.1056855597667566e-09,
+      "loss": 0.0008,
+      "step": 24514
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0042220556015788e-09,
+      "loss": 0.0002,
+      "step": 24515
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.9052638010941255e-09,
+      "loss": 0.0007,
+      "step": 24516
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.808810797909732e-09,
+      "loss": 0.0014,
+      "step": 24517
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7148630476637727e-09,
+      "loss": 0.0101,
+      "step": 24518
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6234205519216615e-09,
+      "loss": 0.0012,
+      "step": 24519
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5344833122155064e-09,
+      "loss": 0.001,
+      "step": 24520
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.448051330010802e-09,
+      "loss": 0.0008,
+      "step": 24521
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.364124606789696e-09,
+      "loss": 0.0002,
+      "step": 24522
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2827031439177626e-09,
+      "loss": 0.0008,
+      "step": 24523
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2037869427605762e-09,
+      "loss": 0.0039,
+      "step": 24524
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1273760046504043e-09,
+      "loss": 0.0054,
+      "step": 24525
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0534703308529013e-09,
+      "loss": 0.0021,
+      "step": 24526
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.820699226004147e-10,
+      "loss": 0.0027,
+      "step": 24527
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.131747811086387e-10,
+      "loss": 0.0003,
+      "step": 24528
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.467849074933474e-10,
+      "loss": 0.0001,
+      "step": 24529
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.829003028869684e-10,
+      "loss": 0.0032,
+      "step": 24530
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.215209683386624e-10,
+      "loss": 0.001,
+      "step": 24531
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.626469048975902e-10,
+      "loss": 0.001,
+      "step": 24532
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.062781135296457e-10,
+      "loss": 0.0072,
+      "step": 24533
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.524145951840698e-10,
+      "loss": 0.0001,
+      "step": 24534
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.010563507601428e-10,
+      "loss": 0.0006,
+      "step": 24535
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.5220338110718566e-10,
+      "loss": 0.0007,
+      "step": 24536
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.058556870412122e-10,
+      "loss": 0.0055,
+      "step": 24537
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.620132693449296e-10,
+      "loss": 0.0003,
+      "step": 24538
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.206761287510851e-10,
+      "loss": 0.0002,
+      "step": 24539
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.8184426594246576e-10,
+      "loss": 0.0051,
+      "step": 24540
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4551768158520557e-10,
+      "loss": 0.0038,
+      "step": 24541
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.1169637626217151e-10,
+      "loss": 0.0026,
+      "step": 24542
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.8038035053957734e-10,
+      "loss": 0.0,
+      "step": 24543
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5156960496698345e-10,
+      "loss": 0.0008,
+      "step": 24544
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2526413999403019e-10,
+      "loss": 0.0001,
+      "step": 24545
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0146395607035784e-10,
+      "loss": 0.0039,
+      "step": 24546
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.016905361230009e-11,
+      "loss": 0.0028,
+      "step": 24547
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.13794329529238e-11,
+      "loss": 0.0007,
+      "step": 24548
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.509509440864256e-11,
+      "loss": 0.0009,
+      "step": 24549
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.131603826256324e-11,
+      "loss": 0.0013,
+      "step": 24550
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.004226474783266e-11,
+      "loss": 0.0001,
+      "step": 24551
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.1273774030984284e-11,
+      "loss": 0.0021,
+      "step": 24552
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.010566261898219e-12,
+      "loss": 0.0004,
+      "step": 24553
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2526415738012274e-12,
+      "loss": 0.0001,
+      "step": 24554
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.0014,
+      "step": 24555
+    },
+    {
+      "epoch": 5.0,
+      "step": 24555,
+      "total_flos": 1.9137197607726612e+18,
+      "train_loss": 0.013171266583756296,
+      "train_runtime": 80458.8335,
+      "train_samples_per_second": 4.883,
+      "train_steps_per_second": 0.305
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 24555,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 1000,
+  "total_flos": 1.9137197607726612e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}